Jupyter Notebook: height statistics of the best competition climbers

This is a static view of the Jupyter notebook I used for my calculations. Here you can download all the source files, I used Python 3.11. Some lines below are cut due to the small width of the blog, if you want you can open the html file from the source files directly in your browser.

Initial imports and a function downloading height¶

In [1]:

%matplotlib inline
import bs4
import requests
import pandas as pd
import numpy as np

from matplotlib import pyplot as plt
from pathlib import Path
from scipy import stats


def get_height(personal_url):
    height = np.nan
    
    html_content = requests.get(personal_url).text

    soup = bs4.BeautifulSoup(html_content, "lxml")

    info = soup.find('div', class_="personal-info")
    if info is not None:
        exists = False
        for p in info.find_all('p'):
            if 'HEIGHT' in p.string:
                exists = True
                continue
            elif exists == True:
                height = int(p.string)
                break
        return height
    
# these links work as per November 2023, they might have changed though
anraku_url = 'https://www.ifsc-climbing.org/index.php?option=com_ifsc&task=athlete.display&id=13040'
song_url = 'https://www.ifsc-climbing.org/index.php?option=com_ifsc&task=athlete.display&id=14018'
hanke_url = 'https://www.ifsc-climbing.org/index.php?option=com_ifsc&task=athlete.display&id=2014'
martin_url = 'https://www.ifsc-climbing.org/index.php?option=com_ifsc&task=athlete.display&id=531'

assert get_height(anraku_url) == 168
assert get_height(song_url) == 168
assert get_height(hanke_url) == 167
assert get_height(martin_url) is np.nan
print('No errors downloading heights')

# rankings were downloaded as html tables from here:
# https://www.ifsc-climbing.org/index.php/world-competition/ranking

No errors downloading heights

Preparing a dataframe for the world ranking¶

In [2]:

data_path = Path('ifsc_data/worldranking.csv')
if data_path.is_file():
    # if the csv alredy exists we can skip the whole process
    df = pd.read_csv(data_path)
else:
    df = pd.DataFrame(columns=['sex', 'full_name', 'height', 'country', 'rank'])

    for sex in ['M', 'F']:
        gender = "men" if sex == "M" else "women"          
        with open(f'ifsc_data/{gender}_worldranking.html', 'r') as file:
            table = bs4.BeautifulSoup(file, 'lxml')
            rows = table.find('tbody').find_all('tr')

        for row in rows:
            df_row = {'sex': sex}
            cols = row.find_all('td')

            df_row['rank'] = int(cols[0].find('p').string)
            df_row['country'] = cols[3].find('a').string

            col_one = cols[1].find('a')
            col_two  = cols[2].find('a')
            df_row['full_name'] = ' '.join([col_one.string.lower(), col_two.string.lower()])

            existing_rows = df.loc[df['full_name'] == df_row['full_name']]
            if len(existing_rows) > 0:
                df_row['height'] = existing_rows.iloc[0]['height']
                print('  Using saved height for', df_row['full_name'], ':', df_row['height'])
            else:
                if col_one.get('href') != col_two.get('href'):
                    print(f'WARNING: Two different urls for {full_name}')
                url = col_one.get('href')
                print('  Downloading height for', df_row['full_name'])
                df_row['height'] = get_height(url)

            df.loc[len(df)] = df_row  

                
df.to_csv(data_path, index=False)

In [3]:

df_rank = pd.read_csv('ifsc_data/worldranking.csv')
print('No. of duplicates:', np.count_nonzero(df_rank.duplicated()))  # just to be sure
df_rank.info()

No. of duplicates: 0
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 218 entries, 0 to 217
Data columns (total 5 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   sex        218 non-null    object 
 1   full_name  218 non-null    object 
 2   height     148 non-null    float64
 3   country    218 non-null    object 
 4   rank       218 non-null    int64  
dtypes: float64(1), int64(1), object(3)
memory usage: 8.6+ KB

Analysis of the world ranking data for men¶

In [4]:

df_m = df_rank.loc[df['sex'] == 'M']
df_m.info()

<class 'pandas.core.frame.DataFrame'>
Index: 106 entries, 0 to 105
Data columns (total 5 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   sex        106 non-null    object 
 1   full_name  106 non-null    object 
 2   height     72 non-null     float64
 3   country    106 non-null    object 
 4   rank       106 non-null    int64  
dtypes: float64(1), int64(1), object(3)
memory usage: 5.0+ KB

In [5]:

df_m.describe()

Out[5]:

	height	rank
count	72.000000	106.000000
mean	174.847222	53.481132
std	6.368186	30.712563
min	162.000000	1.000000
25%	170.000000	27.250000
50%	175.000000	53.500000
75%	179.000000	79.750000
max	198.000000	106.000000

In [6]:

df_m.loc[df_m['height'] > 184]

Out[6]:

	sex	full_name	height	country	rank
9	M	adam ondra	186.0	CZE	10
11	M	meichi narasaki	188.0	JPN	12
15	M	paul jenft	198.0	FRA	16

In [7]:

df_m.loc[df_m['height'] < 165]

Out[7]:

	sex	full_name	height	country	rank
13	M	sascha lehmann	164.0	SUI	14
14	M	sean bailey	163.0	USA	15
19	M	shion omata	162.0	JPN	20
52	M	dillon countryman	164.0	USA	53

In [8]:

mean = df_m['height'].mean()
std = df_m['height'].std()
bins = np.arange(np.floor(df_m['height'].min()), df_m['height'].max()+1, 1)

plt.hist(df_m['height'], bins=bins, color='deepskyblue', zorder=10, alpha=0.9)
plt.axvline(mean, color='black', lw=0.9)
plt.axvline(mean-std, linestyle='--', color='black', lw=0.9)
plt.axvline(mean+std, linestyle='--', color='black', lw=0.9)
ax = plt.gca()
ax.set_xticks(bins[::3])
plt.text(0.6, 0.93, f'Avg. height is {round(mean, 1)} $\pm$ {round(std, 1)}', transform=ax.transAxes)
plt.title('Height distribution of male climbers\nregistered in the IFSC World Ranking 2023')
plt.xlabel('height [cm]')
plt.ylabel('number of climbers')
plt.show()

In [9]:

null_m = df_m.loc[df_m['height'].isnull()]
not_null_m = df_m.loc[df_m['height'].notnull()]

avg_rank = round(df_m['rank'].mean())
avg_rank_nan = round(null_m['rank'].mean())
avg_rank_not_nan = round(not_null_m['rank'].mean())

print(f'Climbers in ranking: {len(null_m)+len(not_null_m)}; with height {len(not_null_m)}, without {len(null_m)}')
print(f'avg. rank: {avg_rank}; avg. rank of climbers w/out height: {avg_rank_nan},'+
      f' and with height data: {avg_rank_not_nan}')

Climbers in ranking: 106; with height 72, without 34
avg. rank: 53; avg. rank of climbers w/out height: 82, and with height data: 40

Statistical tests for the distribution and comparison with the general population¶

I perform 3 tests here, which are repeated for other histograms as well. The first one is D’Agostino and Pearson’s test for normality to see if the distribution could be Gaussian. The second one is Kolmogorov-Smirnov test comparing the empirical distribution with a normal distribution of mean and standard deviation corresponding to those of the general population. This test is to see whether the best climbers could have the same height distribution as the whole society. The last test is Student's t-test comparing just the average value for climbers with the general population (for this test the assumption is that the distribution is normal, hence the first test).

In [10]:

# Paul Jenft is 10cm taller than the next tallest climber,
# so we can take him out as an outlier
df_m_clean = df_m.loc[(df_m['height'].notnull()) & (df_m['height'] < 195)]
stats.normaltest(df_m_clean['height'])

Out[10]:

NormaltestResult(statistic=1.1058087166129469, pvalue=0.5752765724237572)

In [11]:

world_m_height_avg = 178.4  # source: https://doi.org/10.7554/eLife.20320.001
world_m_height_std = 7.59   # it's an average for men born between 1980 and 1994

stats.ks_1samp(df_m_clean['height'], stats.norm.cdf, args=(world_m_height_avg, world_m_height_std))

Out[11]:

KstestResult(statistic=0.28539728488157, pvalue=1.3022112549577226e-05, statistic_location=179.0, statistic_sign=1)

In [12]:

stats.ttest_1samp(df_m_clean['height'], world_m_height_avg)

Out[12]:

TtestResult(statistic=-5.658138174840393, pvalue=3.107588094982158e-07, df=70)

Statisticts for the top 50 climbers¶

In [13]:

df_m_top_50 = df_m.loc[df['rank'] <= 50]
df_m_top_50.info()

<class 'pandas.core.frame.DataFrame'>
Index: 50 entries, 0 to 49
Data columns (total 5 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   sex        50 non-null     object 
 1   full_name  50 non-null     object 
 2   height     49 non-null     float64
 3   country    50 non-null     object 
 4   rank       50 non-null     int64  
dtypes: float64(1), int64(1), object(3)
memory usage: 2.3+ KB

In [14]:

df_m_top_50.describe() # the average of top 50 is 174cm without Paul Jenft

Out[14]:

	height	rank
count	49.000000	50.00000
mean	174.653061	25.50000
std	6.796295	14.57738
min	162.000000	1.00000
25%	170.000000	13.25000
50%	175.000000	25.50000
75%	178.000000	37.75000
max	198.000000	50.00000

In [15]:

mean = df_m_top_50['height'].mean()
std = df_m_top_50['height'].std()
bins = np.arange(np.floor(df_m_top_50['height'].min()), df_m_top_50['height'].max()+1, 1)

plt.hist(df_m_top_50['height'], bins=bins, color='deepskyblue', zorder=10, alpha=0.9)
plt.axvline(mean, color='black', lw=0.9)
plt.axvline(mean-std, linestyle='--', color='black', lw=0.9)
plt.axvline(mean+std, linestyle='--', color='black', lw=0.9)
ax = plt.gca()
ax.set_xticks(bins[::3])
plt.text(0.6, 0.93, f'Avg. height is {round(mean, 1)} $\pm$ {round(std, 1)}', transform=ax.transAxes)
plt.title('Height distribution of the top 50 male climbers')
plt.xlabel('height [cm]')
plt.ylabel('number of climbers')
plt.show()

In [16]:

df_m_clean_50 = df_m_top_50.loc[(df_m_top_50['height'].notnull()) & (df_m_top_50['height'] < 195)]
stats.normaltest(df_m_clean_50['height'])

Out[16]:

NormaltestResult(statistic=0.22144894346920682, pvalue=0.8951853638232864)

In [17]:

stats.ks_1samp(df_m_clean_50['height'], stats.norm.cdf, args=(world_m_height_avg, world_m_height_std))

Out[17]:

KstestResult(statistic=0.3226625430975324, pvalue=6.0124285907116906e-05, statistic_location=179.0, statistic_sign=1)

In [18]:

stats.ttest_1samp(df_m_clean_50['height'], world_m_height_avg)

Out[18]:

TtestResult(statistic=-4.93413218077492, pvalue=1.050704076544895e-05, df=47)

Countries with the most ranked climbers and their average height¶

In [19]:

df_m_country = df_m.groupby(['country'], as_index=False).agg({'full_name': 'count', 'height': 'mean'})
df_m_country = df_m_country.sort_values('full_name', ascending=False)
df_m_country.rename(columns={'full_name': 'no. of climbers', 'height': 'avg. height'})

Out[19]:

	country	no. of climbers	avg. height
21	ITA	7	174.800000
12	GBR	7	175.166667
35	USA	6	171.000000
2	AUT	6	178.000000
22	JPN	6	172.600000
30	SLO	6	178.333333
11	FRA	5	179.200000
23	KOR	5	173.000000
6	CAN	5	172.750000
20	ISR	5	168.333333
1	AUS	4	173.500000
13	GER	3	176.000000
4	BRA	3	NaN
16	INA	3	171.000000
3	BEL	3	174.000000
9	CZE	3	182.000000
26	MEX	2	NaN
31	SUI	2	169.500000
33	SWE	2	177.000000
18	IRI	2	NaN
14	HKG	2	172.000000
10	ESP	2	181.000000
8	CHN	2	173.000000
7	CHI	2	180.000000
5	BUL	2	172.000000
19	IRL	1	174.000000
17	IND	1	NaN
15	HUN	1	NaN
24	LAT	1	181.000000
25	LUX	1	NaN
27	PER	1	NaN
28	RSA	1	174.000000
29	SGP	1	NaN
32	SVK	1	175.000000
34	THA	1	NaN
0	ARG	1	NaN

Analysis of the world ranking data for women¶

In [20]:

df_f = df_rank.loc[df['sex'] == 'F']
df_f.info()

<class 'pandas.core.frame.DataFrame'>
Index: 112 entries, 106 to 217
Data columns (total 5 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   sex        112 non-null    object 
 1   full_name  112 non-null    object 
 2   height     76 non-null     float64
 3   country    112 non-null    object 
 4   rank       112 non-null    int64  
dtypes: float64(1), int64(1), object(3)
memory usage: 5.2+ KB

In [21]:

df_f.describe()

Out[21]:

	height	rank
count	76.000000	112.000000
mean	163.289474	56.410714
std	5.285996	32.394681
min	150.000000	1.000000
25%	160.000000	28.750000
50%	163.000000	56.500000
75%	166.000000	84.250000
max	175.000000	112.000000

In [22]:

df_f.loc[df_f['height'] > 173]

Out[22]:

	sex	full_name	height	country	rank
125	F	stasa gejo	175.0	SRB	20
143	F	flavy cohaut	174.0	FRA	38
146	F	julija kruder	175.0	SLO	41

In [23]:

df_f.loc[df_f['height'] < 153]

Out[23]:

	sex	full_name	height	country	rank
114	F	jain kim	152.0	KOR	9
128	F	laura rogora	152.0	ITA	23
181	F	chaeyeong kim	150.0	KOR	75

In [24]:

mean = df_f['height'].mean()
std = df_f['height'].std()
bins = np.arange(np.floor(df_f['height'].min()), df_f['height'].max()+1, 1)

plt.hist(df_f['height'], bins=bins, color='tomato', zorder=10, alpha=0.9)
plt.axvline(mean, color='black', lw=0.9)
plt.axvline(mean-std, linestyle='--', color='black', lw=0.9)
plt.axvline(mean+std, linestyle='--', color='black', lw=0.9)
ax = plt.gca()
ax.set_xticks(bins[::3])
plt.text(0.05, 0.87, f'Avg. height\nis {round(mean, 1)} $\pm$ {round(std, 1)}', transform=ax.transAxes)
plt.title('Height distribution of female climbers\nregistered in the IFSC World Ranking 2023')
plt.xlabel('height [cm]')
plt.ylabel('number of climbers')
plt.show()

In [25]:

null_f = df_f.loc[df_f['height'].isnull()]
not_null_f = df_f.loc[df_f['height'].notnull()]

avg_rank = round(df_f['rank'].mean())
avg_rank_nan = round(null_f['rank'].mean())
avg_rank_not_nan = round(not_null_f['rank'].mean())

print(f'Climbers in ranking: {len(null_f)+len(not_null_f)}; with height {len(not_null_f)}, without {len(null_f)}')
print(f'avg. rank: {avg_rank}; avg. rank of climbers w/out height: {avg_rank_nan},'+
      f' and with height data: {avg_rank_not_nan}')

Climbers in ranking: 112; with height 76, without 36
avg. rank: 56; avg. rank of climbers w/out height: 87, and with height data: 42

Statistical tests for the distribution and comparison with the general population¶

In [26]:

df_f_clean = df_f.loc[df_f['height'].notnull()]
stats.normaltest(df_f_clean['height'])

Out[26]:

NormaltestResult(statistic=0.34554396819934946, pvalue=0.8413294294627298)

In [27]:

world_f_height_avg = 164.7  # source: https://doi.org/10.7554/eLife.20320.001
world_f_height_std = 7.07   # it's an average for women born between 1980 and 1994

stats.ks_1samp(df_f_clean['height'], stats.norm.cdf, args=(world_f_height_avg, world_f_height_std))

Out[27]:

KstestResult(statistic=0.21991894003448065, pvalue=0.0010427677664592615, statistic_location=165.0, statistic_sign=1)

In [28]:

stats.ttest_1samp(df_f_clean['height'], world_f_height_avg)

Out[28]:

TtestResult(statistic=-2.3262755376972337, pvalue=0.02270781036453741, df=75)

Statisticts for the top 50 climbers¶

In [29]:

df_f_top_50 = df_f.loc[df['rank'] <= 50]
df_f_top_50.info()

<class 'pandas.core.frame.DataFrame'>
Index: 50 entries, 106 to 155
Data columns (total 5 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   sex        50 non-null     object 
 1   full_name  50 non-null     object 
 2   height     50 non-null     float64
 3   country    50 non-null     object 
 4   rank       50 non-null     int64  
dtypes: float64(1), int64(1), object(3)
memory usage: 2.3+ KB

In [30]:

df_f_top_50.describe()

Out[30]:

	height	rank
count	50.000000	50.000000
mean	163.580000	25.480000
std	5.496158	14.578766
min	152.000000	1.000000
25%	160.000000	13.250000
50%	163.000000	25.000000
75%	166.750000	37.750000
max	175.000000	50.000000

In [31]:

mean = df_f_top_50['height'].mean()
std = df_f_top_50['height'].std()
bins = np.arange(np.floor(df_f_top_50['height'].min()), df_f_top_50['height'].max()+1, 1)

plt.hist(df_f_top_50['height'], bins=bins, color='tomato', zorder=10, alpha=0.9)
plt.axvline(mean, color='black', lw=0.9)
plt.axvline(mean-std, linestyle='--', color='black', lw=0.9)
plt.axvline(mean+std, linestyle='--', color='black', lw=0.9)
ax = plt.gca()
ax.set_xticks(bins[::3])
plt.text(0.05, 0.87, f'Avg. height\nis {round(mean, 1)} $\pm$ {round(std, 1)}', transform=ax.transAxes)
plt.title('Height distribution of the top 50 female climbers')
plt.xlabel('height [cm]')
plt.ylabel('number of climbers')
plt.show()

In [32]:

df_f_clean_50 = df_f_top_50.loc[df_f_top_50['height'].notnull()]
stats.normaltest(df_f_clean_50['height'])

Out[32]:

NormaltestResult(statistic=0.3251394770115752, pvalue=0.8499568134403317)

In [33]:

stats.ks_1samp(df_f_clean_50['height'], stats.norm.cdf, args=(world_f_height_avg, world_f_height_std))

Out[33]:

KstestResult(statistic=0.20307683477132277, pvalue=0.02766478274130235, statistic_location=165.0, statistic_sign=1)

In [34]:

stats.ttest_1samp(df_f_clean_50['height'], world_f_height_avg)

Out[34]:

TtestResult(statistic=-1.4409330376615785, pvalue=0.15596368735799032, df=49)

Countries with the most ranked climbers and their average height¶

In [35]:

df_f_country = df_f.groupby(['country'], as_index=False).agg({'full_name': 'count', 'height': 'mean'})
df_f_country = df_f_country.sort_values('full_name', ascending=False)
df_f_country.rename(columns={'full_name': 'no. of climbers', 'height': 'avg. height'})

Out[35]:

	country	no. of climbers	avg. height
38	USA	9	162.833333
32	SLO	8	167.125000
13	FRA	7	163.714286
22	JPN	7	162.571429
15	GER	5	164.750000
24	KOR	5	156.250000
10	CZE	4	162.000000
14	GBR	4	161.333333
20	ISR	4	166.500000
7	CHI	4	160.666667
6	CAN	4	161.000000
4	BRA	4	164.000000
21	ITA	4	159.000000
25	MEX	3	NaN
1	AUS	3	173.000000
34	SUI	3	163.333333
37	UKR	3	159.000000
0	ARG	3	157.000000
2	AUT	3	165.000000
8	CHN	3	160.000000
18	IRI	2	161.000000
11	ESP	2	NaN
30	PUR	1	NaN
3	BEL	1	166.000000
36	TPE	1	170.000000
35	SVK	1	169.000000
5	BUL	1	162.000000
33	SRB	1	175.000000
31	RSA	1	NaN
29	PER	1	168.000000
19	ISL	1	171.000000
28	NOR	1	NaN
27	NED	1	168.000000
26	MKD	1	NaN
9	CRO	1	NaN
23	KAZ	1	NaN
12	FIN	1	NaN
16	INA	1	158.000000
17	IND	1	NaN
39	VEN	1	NaN

Preparing a dataframe for the world cups¶

In [36]:

data_path = Path('ifsc_data/world_cups.csv')
if data_path.is_file():
    df = pd.read_csv(data_path)
else:
    df = pd.DataFrame(columns=['year', 'category', 'sex', 'full_name', 'height', 'country', 'rank'])

years = list(range(2013, 2024))[::-1]
for year in years:
    if len(df.loc[df['year'] == year]) > 0 or year == 2020:
        # I assume that all the data for this year has been loeaded before
        continue
    for sex in ['M', 'F']:
        for cat in ['lead', 'boulder']:
            gender = "men" if sex == "M" else "women"
            print(f"Loading year {year} for {gender}'s {cat}")
            
            with open(f'ifsc_data/{gender}_{cat}_{year}.html', 'r') as file:
                table = bs4.BeautifulSoup(file, 'lxml')
                rows = table.find('tbody').find_all('tr')

            for row in rows:
                df_row = {'year': year, 'category': cat, 'sex': sex}
                cols = row.find_all('td')
                
                df_row['rank'] = int(cols[0].find('p').string)
                df_row['country'] = cols[3].find('a').string
                
                col_one = cols[1].find('a')
                col_two  = cols[2].find('a')
                df_row['full_name'] = ' '.join([col_one.string.lower(), col_two.string.lower()])
                
                existing_rows = df.loc[df['full_name'] == df_row['full_name']]
                if len(existing_rows) > 0:
                    df_row['height'] = existing_rows.iloc[0]['height']
                    print('  Using saved height for', df_row['full_name'], ':', df_row['height'])
                else:
                    if col_one.get('href') != col_two.get('href'):
                        print(f'WARNING: Two different urls for {full_name}')
                    url = col_one.get('href')
                    df_row['height'] = get_height(url)
                
                df.loc[len(df)] = df_row  

                
df.to_csv(data_path, index=False)

In [37]:

df_cup = pd.read_csv('ifsc_data/world_cups.csv')
print('No. of duplicates:', np.count_nonzero(df_cup.duplicated()))  # just to be sure
df_cup.info()

No. of duplicates: 0
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 4358 entries, 0 to 4357
Data columns (total 7 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   year       4358 non-null   int64  
 1   category   4358 non-null   object 
 2   sex        4358 non-null   object 
 3   full_name  4358 non-null   object 
 4   height     1849 non-null   float64
 5   country    4358 non-null   object 
 6   rank       4358 non-null   int64  
dtypes: float64(1), int64(2), object(4)
memory usage: 238.5+ KB

Basic statistics for both categories for men¶

In [38]:

df_cup_m = df_cup.loc[df_cup['sex'] == 'M']
df_cup_m.info()

<class 'pandas.core.frame.DataFrame'>
Index: 2186 entries, 0 to 4143
Data columns (total 7 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   year       2186 non-null   int64  
 1   category   2186 non-null   object 
 2   sex        2186 non-null   object 
 3   full_name  2186 non-null   object 
 4   height     921 non-null    float64
 5   country    2186 non-null   object 
 6   rank       2186 non-null   int64  
dtypes: float64(1), int64(2), object(4)
memory usage: 136.6+ KB

In [39]:

column = df_cup_m.copy()['category']
df_cup_m.insert(0, 'category_dup', column, allow_duplicates=False)
df_cup_m_agg = df_cup_m.groupby(['full_name'], as_index=False)
df_cup_m_agg = df_cup_m_agg.agg({'full_name': 'first', 'country': 'first', 'year': 'count',
                                 'category': lambda x: (x == 'lead').sum(),
                                 'category_dup': lambda x: (x == 'boulder').sum(),
                                 'height': 'mean', 'rank': 'mean'})
df_cup_m_agg = df_cup_m_agg.sort_values('year', ascending=False)
df_cup_m_agg = df_cup_m_agg.rename(columns={'year': 'times in ranking', 'rank' : 'avg. rank',
                                            'category': 'in lead', 'category_dup': 'in boulder'})

df_cup_m_agg.info()

<class 'pandas.core.frame.DataFrame'>
Index: 705 entries, 256 to 704
Data columns (total 7 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   full_name         705 non-null    object 
 1   country           705 non-null    object 
 2   times in ranking  705 non-null    int64  
 3   in lead           705 non-null    int64  
 4   in boulder        705 non-null    int64  
 5   height            171 non-null    float64
 6   avg. rank         705 non-null    float64
dtypes: float64(2), int64(3), object(2)
memory usage: 44.1+ KB

In [40]:

df_cup_m_agg.head(20)

Out[40]:

	full_name	country	times in ranking	in lead	in boulder	height	avg. rank
256	jakob schubert	AUT	19	10	9	176.0	12.157895
550	sean mccoll	CAN	18	8	10	169.0	23.277778
345	kokoro fujii	JPN	17	7	10	176.0	16.764706
150	domen skofic	SLO	15	9	6	177.0	28.933333
292	jongwon chon	KOR	15	6	9	177.0	29.933333
384	marcello bombardi	ITA	15	10	5	177.0	42.333333
424	michael piccolruaz	ITA	15	5	10	178.0	45.066667
222	hannes puman	SWE	15	10	5	177.0	47.333333
637	tomoa narasaki	JPN	15	6	9	170.0	17.000000
2	adam ondra	CZE	15	10	5	186.0	14.800000
271	jernej kruder	SLO	14	4	10	179.0	35.214286
548	sean bailey	USA	13	7	6	163.0	17.923077
672	yannick flohé	GER	13	7	6	178.0	30.461538
546	sascha lehmann	SUI	13	8	5	164.0	30.384615
461	nicolas collin	BEL	13	8	5	179.0	41.923077
678	yoshiyuki ogata	JPN	13	4	9	172.0	18.230769
16	alex khazanov	ISR	13	4	9	NaN	58.923077
266	jan hojer	GER	13	5	8	188.0	21.769231
571	simon lorenzi	BEL	12	7	5	168.0	50.500000
591	stefano ghisolfi	ITA	12	10	2	169.0	15.916667

In [41]:

df_cup_m_agg.describe()

Out[41]:

	times in ranking	in lead	in boulder	height	avg. rank
count	705.000000	705.000000	705.000000	171.000000	705.000000
mean	3.100709	1.506383	1.594326	174.157895	74.807740
std	3.114722	2.024783	1.967050	6.184771	33.989163
min	1.000000	0.000000	0.000000	155.000000	1.000000
25%	1.000000	0.000000	0.000000	170.000000	50.000000
50%	2.000000	1.000000	1.000000	175.000000	73.666667
75%	4.000000	2.000000	2.000000	178.000000	97.000000
max	19.000000	10.000000	10.000000	198.000000	160.000000

In [42]:

df_cup_m_agg_country = df_cup_m_agg.groupby(['country'], as_index=False)
df_cup_m_agg_country = df_cup_m_agg_country.agg({'full_name': 'count', 'height': 'mean'})
df_cup_m_agg_country = df_cup_m_agg_country.sort_values('full_name', ascending=False)
df_cup_m_agg_country.rename(columns={'full_name': 'no. of climbers', 'height': 'avg. height'})
df_cup_m_agg_country.head(30)

Out[42]:

	country	full_name	height
27	JPN	62	172.090909
15	FRA	57	176.687500
58	USA	56	175.533333
18	GER	32	176.200000
7	CAN	29	172.750000
29	KOR	29	170.250000
47	RUS	28	174.500000
2	AUT	28	178.428571
16	GBR	26	175.166667
51	SUI	23	169.571429
26	ITA	23	171.625000
49	SLO	21	176.363636
9	CHN	20	173.000000
13	ESP	17	178.250000
21	INA	15	168.500000
1	AUS	14	176.000000
23	IRI	14	173.000000
10	CZE	12	178.333333
4	BEL	12	174.750000
35	MEX	12	NaN
22	IND	12	NaN
25	ISR	11	171.250000
19	HKG	11	172.000000
39	NOR	9	175.000000
57	UKR	9	179.000000
42	POL	9	170.500000
37	NED	9	173.500000
8	CHI	8	180.000000
48	SGP	7	NaN
53	SWE	7	177.000000

Analysis of the lead world cup data for men¶

In [43]:

df_m_lead = df_cup.loc[(df_cup['sex'] == 'M') & (df_cup['category'] == 'lead')]
df_m_lead.info()

<class 'pandas.core.frame.DataFrame'>
Index: 1062 entries, 0 to 4028
Data columns (total 7 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   year       1062 non-null   int64  
 1   category   1062 non-null   object 
 2   sex        1062 non-null   object 
 3   full_name  1062 non-null   object 
 4   height     490 non-null    float64
 5   country    1062 non-null   object 
 6   rank       1062 non-null   int64  
dtypes: float64(1), int64(2), object(4)
memory usage: 66.4+ KB

In [44]:

df_m_lead.describe()
# note that in the numbers below a climber who was ranked for several years
# is taken into account that many times in the calculations

Out[44]:

	year	height	rank
count	1062.000000	490.000000	1062.000000
mean	2018.275895	174.181633	56.488701
std	3.440920	6.616710	36.212970
min	2013.000000	155.000000	1.000000
25%	2015.000000	169.250000	27.000000
50%	2018.000000	175.000000	53.500000
75%	2022.000000	178.000000	80.000000
max	2023.000000	198.000000	160.000000

Analysis of aggregated lead rankings for all years¶

In [45]:

df_m_lead_agg = df_m_lead.groupby(['full_name'], as_index=False)
df_m_lead_agg = df_m_lead_agg.agg({'full_name': 'first', 'country': 'first', 'year': 'count',
                                   'height': 'mean', 'rank': 'mean'})
df_m_lead_agg = df_m_lead_agg.sort_values('year', ascending=False)
df_m_lead_agg = df_m_lead_agg.rename(columns={'year': 'years in ranking', 'rank' : 'avg. rank'})

df_m_lead_agg.info()

<class 'pandas.core.frame.DataFrame'>
Index: 431 entries, 0 to 430
Data columns (total 5 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   full_name         431 non-null    object 
 1   country           431 non-null    object 
 2   years in ranking  431 non-null    int64  
 3   height            140 non-null    float64
 4   avg. rank         431 non-null    float64
dtypes: float64(2), int64(1), object(2)
memory usage: 20.2+ KB

In [46]:

df_m_lead_agg.head(20)

Out[46]:

	full_name	country	years in ranking	height	avg. rank
0	adam ondra	CZE	10	186.0	17.000000
237	martin bergant	SLO	10	182.0	44.200000
335	sebastian halenke	GER	10	177.0	19.100000
232	marcello bombardi	ITA	10	177.0	24.900000
358	stefano ghisolfi	ITA	10	169.0	12.000000
131	hannes puman	SWE	10	177.0	34.900000
154	jakob schubert	AUT	10	176.0	8.800000
86	domen skofic	SLO	9	177.0	8.111111
61	christoph hanke	GER	9	167.0	39.222222
262	milan preskar	SLO	8	172.0	58.625000
330	sascha lehmann	SUI	8	164.0	17.500000
133	hanwool kim	KOR	8	NaN	34.750000
81	dimitri vogt	SUI	8	158.0	75.375000
283	nicolas collin	BEL	8	179.0	42.875000
334	sean mccoll	CAN	8	169.0	15.875000
244	masahiro higuchi	JPN	8	169.0	14.375000
250	max rudigier	AUT	7	NaN	35.285714
273	nao monchois	FRA	7	NaN	52.000000
209	kokoro fujii	JPN	7	176.0	24.285714
246	mathias posch	AUT	7	171.0	59.142857

In [47]:

df_m_lead_agg.describe()

Out[47]:

	years in ranking	height	avg. rank
count	431.000000	140.000000	431.000000
mean	2.464037	173.964286	71.940380
std	2.084936	6.441112	35.360244
min	1.000000	155.000000	1.000000
25%	1.000000	170.000000	47.750000
50%	2.000000	175.000000	69.000000
75%	3.000000	178.000000	90.750000
max	10.000000	198.000000	160.000000

In [48]:

df_m_lead_agg.loc[df_m_lead_agg['years in ranking'] >= 5].describe()

Out[48]:

	years in ranking	height	avg. rank
count	63.000000	38.000000	63.000000
mean	6.746032	173.605263	36.988946
std	1.585908	6.499644	18.582166
min	5.000000	158.000000	6.857143
25%	5.000000	169.000000	20.557143
50%	7.000000	174.000000	37.142857
75%	7.500000	177.750000	51.700000
max	10.000000	188.000000	75.375000

In [49]:

df_m_lead_agg.loc[df_m_lead_agg['height'] > 186]

Out[49]:

	full_name	country	years in ranking	height	avg. rank
159	jan hojer	GER	5	188.0	42.40
295	paul jenft	FRA	4	198.0	40.25
253	meichi narasaki	JPN	4	188.0	36.50
215	louis gundolf	AUT	4	188.0	76.75
62	christoph schweiger	GER	1	187.0	117.00

In [50]:

df_m_lead_agg.loc[df_m_lead_agg['height'] < 163]

Out[50]:

	full_name	country	years in ranking	height	avg. rank
81	dimitri vogt	SUI	8	158.0	75.375
146	hyunbin min	KOR	4	162.0	23.000
121	giovanni placci	ITA	2	155.0	42.000
346	shion omata	JPN	1	162.0	4.000
400	veddriq leonardo	INA	1	162.0	84.000

In [51]:

mean = df_m_lead_agg['height'].mean()
std = df_m_lead_agg['height'].std()
bins = np.arange(np.floor(df_m_lead_agg['height'].min()), df_m_lead_agg['height'].max()+1, 1)

plt.hist(df_m_lead_agg['height'], bins=bins, color='deepskyblue', zorder=10, alpha=0.9)
plt.axvline(mean, color='black', lw=0.9)
plt.axvline(mean-std, linestyle='--', color='black', lw=0.9)
plt.axvline(mean+std, linestyle='--', color='black', lw=0.9)
ax = plt.gca()
ax.set_xticks(bins[::3])
plt.text(0.61, 0.93, f'Avg. height is {round(mean, 1)} $\pm$ {round(std, 1)}', transform=ax.transAxes)
plt.title('Height distribution of male climbers registered in\nthe IFSC '+
          'Lead World Cup ranking between 2013 and 2023')
plt.xlabel('height [cm]')
plt.ylabel('number of climbers')
plt.show()

In [52]:

df_m_lead_agg_clean = df_m_lead_agg.loc[(df_m_lead_agg['height'].notnull())
                                        & (df_m_lead_agg['height'] < 195) & (df_m_lead_agg['height'] > 156)]
stats.normaltest(df_m_lead_agg_clean['height'])

Out[52]:

NormaltestResult(statistic=0.064157677419489, pvalue=0.968430229279705)

In [53]:

stats.ks_1samp(df_m_lead_agg_clean['height'], stats.norm.cdf, args=(world_m_height_avg, world_m_height_std))

Out[53]:

KstestResult(statistic=0.31086997037047653, pvalue=2.4086640889450246e-12, statistic_location=178.0, statistic_sign=1)

In [54]:

stats.ttest_1samp(df_m_lead_agg_clean['height'], world_m_height_avg)

Out[54]:

TtestResult(statistic=-8.849233874164154, pvalue=3.960169063117888e-15, df=137)

In [55]:

df_m_lead_agg_country = df_m_lead_agg.groupby(['country'], as_index=False)
df_m_lead_agg_country = df_m_lead_agg_country.agg({'full_name': 'count', 'height': 'mean'})
df_m_lead_agg_country = df_m_lead_agg_country.sort_values('full_name', ascending=False)
df_m_lead_agg_country.rename(columns={'full_name': 'no. of climbers', 'height': 'avg. height'})

Out[55]:

	country	no. of climbers	avg. height
23	JPN	40	172.833333
12	FRA	39	176.307692
25	KOR	25	168.333333
47	USA	24	174.833333
37	RUS	23	174.500000
39	SLO	21	176.363636
2	AUT	18	178.428571
22	ITA	17	172.000000
14	GER	16	177.000000
8	CHN	16	173.000000
13	GBR	15	175.166667
17	INA	15	168.500000
41	SUI	13	167.800000
6	CAN	11	172.750000
10	ESP	11	181.000000
15	HKG	9	172.000000
3	BEL	9	174.750000
9	CZE	9	182.000000
30	NOR	8	175.000000
7	CHI	7	180.000000
21	ISR	6	168.333333
1	AUS	6	176.000000
27	MEX	5	NaN
33	POL	5	170.500000
44	THA	5	NaN
28	NED	5	165.000000
18	IND	5	NaN
46	UKR	5	NaN
19	IRI	4	NaN
43	SWE	4	177.000000
4	BRA	4	NaN
16	HUN	3	176.000000
42	SVK	3	175.000000
5	BUL	3	172.000000
45	TPE	2	NaN
0	ARG	2	NaN
48	VEN	2	NaN
36	RSA	2	174.000000
34	POR	2	NaN
31	PAK	2	170.000000
29	NEP	2	NaN
38	SGP	1	NaN
35	ROU	1	NaN
40	SRB	1	175.000000
32	PER	1	NaN
26	LUX	1	NaN
20	IRL	1	174.000000
11	FIN	1	180.000000
24	KAZ	1	170.000000

Changes in the statistics over the years¶

In [56]:

years = list(range(2013, 2024))
years.pop(years.index(2020))
l_height_list = []
l_height_std_list = []

for year in years:
    heights = df_m_lead.loc[df_m_lead['year'] == year]['height']
    l_height_list.append(heights.mean())
    l_height_std_list.append(heights.std())
    
    avg = round(l_height_list[-1], 1)
    std = round(l_height_std_list[-1], 1)
    all_ = len(heights)
    nans = len(heights.loc[heights.isnull()])
    perc = round(100.0*nans/all_, 1)
    print(f'Avg. height {year}: {avg} +- {std} '+
          f'({all_-nans} records, {nans} NaN, {perc}% of NaNs in {all_} climbers)')

Avg. height 2013: 173.8 +- 6.2 (15 records, 82 NaN, 84.5% of NaNs in 97 climbers)
Avg. height 2014: 172.6 +- 6.3 (23 records, 82 NaN, 78.1% of NaNs in 105 climbers)
Avg. height 2015: 173.4 +- 7.4 (23 records, 66 NaN, 74.2% of NaNs in 89 climbers)
Avg. height 2016: 173.2 +- 6.8 (28 records, 64 NaN, 69.6% of NaNs in 92 climbers)
Avg. height 2017: 173.6 +- 6.7 (39 records, 63 NaN, 61.8% of NaNs in 102 climbers)
Avg. height 2018: 173.6 +- 6.8 (47 records, 47 NaN, 50.0% of NaNs in 94 climbers)
Avg. height 2019: 175.1 +- 6.9 (56 records, 27 NaN, 32.5% of NaNs in 83 climbers)
Avg. height 2021: 175.1 +- 6.6 (68 records, 18 NaN, 20.9% of NaNs in 86 climbers)
Avg. height 2022: 174.5 +- 6.4 (97 records, 63 NaN, 39.4% of NaNs in 160 climbers)
Avg. height 2023: 174.2 +- 6.5 (94 records, 60 NaN, 39.0% of NaNs in 154 climbers)

In [57]:

for year in years:
    df_y = df_m_lead.loc[df_m_lead['year'] == year]
    avg_rank = round(df_y['rank'].mean())
    avg_rank_nan = round(df_y.loc[df_y['height'].isnull()]['rank'].mean())
    avg_rank_not_nan = round(df_y.loc[df_y['height'].notnull()]['rank'].mean())
    print(f'{year} avg. rank: {avg_rank}; avg. rank of climbers w/out height {avg_rank_nan},'+
          f' and with height data {avg_rank_not_nan}')

2013 avg. rank: 49; avg. rank of climbers w/out height 52, and with height data 32
2014 avg. rank: 53; avg. rank of climbers w/out height 55, and with height data 43
2015 avg. rank: 45; avg. rank of climbers w/out height 48, and with height data 37
2016 avg. rank: 46; avg. rank of climbers w/out height 50, and with height data 39
2017 avg. rank: 51; avg. rank of climbers w/out height 58, and with height data 40
2018 avg. rank: 47; avg. rank of climbers w/out height 51, and with height data 44
2019 avg. rank: 42; avg. rank of climbers w/out height 48, and with height data 39
2021 avg. rank: 43; avg. rank of climbers w/out height 57, and with height data 40
2022 avg. rank: 80; avg. rank of climbers w/out height 112, and with height data 60
2023 avg. rank: 77; avg. rank of climbers w/out height 110, and with height data 55

In [58]:

outlier_marker = dict(markerfacecolor='none', marker='o', markeredgecolor='orangered', alpha=0.7)
meanprops = dict(markerfacecolor='none', marker='x', markeredgecolor='gray', alpha=0.9, markersize=4)

df_m_lead.boxplot(column='height', by='year', flierprops=outlier_marker, showmeans=True, meanprops=meanprops)
ax = plt.gca()
ax.grid(linestyle='--', alpha=0.5)
ax.grid(axis='x')
ax.set_yticks(np.arange(155, 200, 5))
plt.suptitle('')
plt.title('Box plot of height distribution of male climbers registered\n'+
          'in the IFSC Lead World Cup ranking for different years')
plt.xlabel('year')
plt.ylabel('height [cm]')
plt.show()

Analysis of the bouldring world cup data for men¶

In [59]:

df_m_boulder = df_cup.loc[(df_cup['sex'] == 'M') & (df_cup['category'] == 'boulder')]
df_m_boulder.info()

<class 'pandas.core.frame.DataFrame'>
Index: 1124 entries, 160 to 4143
Data columns (total 7 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   year       1124 non-null   int64  
 1   category   1124 non-null   object 
 2   sex        1124 non-null   object 
 3   full_name  1124 non-null   object 
 4   height     431 non-null    float64
 5   country    1124 non-null   object 
 6   rank       1124 non-null   int64  
dtypes: float64(1), int64(2), object(4)
memory usage: 70.2+ KB

In [60]:

df_m_boulder.describe()
# note that in the numbers below a climber who was ranked for several years
# is taken into account that many times in the calculations

Out[60]:

	year	height	rank
count	1124.000000	431.000000	1124.000000
mean	2017.981317	175.345708	59.982206
std	3.476240	5.775145	37.950331
min	2013.000000	162.000000	1.000000
25%	2015.000000	172.000000	29.000000
50%	2017.000000	176.000000	56.000000
75%	2022.000000	178.000000	85.250000
max	2023.000000	198.000000	149.000000

Analysis of aggregated bouldering rankings for all years¶

In [61]:

df_m_boulder_agg = df_m_boulder.groupby(['full_name'], as_index=False)
df_m_boulder_agg = df_m_boulder_agg.agg({'full_name': 'first', 'country': 'first', 'year': 'count',
                                   'height': 'mean', 'rank': 'mean'})
df_m_boulder_agg = df_m_boulder_agg.sort_values('year', ascending=False)
df_m_boulder_agg = df_m_boulder_agg.rename(columns={'year': 'years in ranking', 'rank' : 'avg. rank'})

df_m_boulder_agg.info()

<class 'pandas.core.frame.DataFrame'>
Index: 482 entries, 232 to 481
Data columns (total 5 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   full_name         482 non-null    object 
 1   country           482 non-null    object 
 2   years in ranking  482 non-null    int64  
 3   height            127 non-null    float64
 4   avg. rank         482 non-null    float64
dtypes: float64(2), int64(1), object(2)
memory usage: 22.6+ KB

In [62]:

df_m_boulder_agg.head(20)

Out[62]:

	full_name	country	years in ranking	height	avg. rank
232	kokoro fujii	JPN	10	176.0	11.500000
287	michael piccolruaz	ITA	10	178.0	36.400000
378	sean mccoll	CAN	10	169.0	29.200000
181	jernej kruder	SLO	10	179.0	18.200000
432	tomoa narasaki	JPN	9	170.0	8.000000
169	jakob schubert	AUT	9	176.0	15.888889
289	mickael mawem	FRA	9	179.0	27.333333
462	yoshiyuki ogata	JPN	9	172.0	19.333333
198	jongwon chon	KOR	9	177.0	7.666667
11	alex khazanov	ISR	9	NaN	40.111111
305	nathaniel coleman	USA	9	182.0	48.444444
349	rei sugimoto	JPN	9	172.0	24.333333
265	martin stranik	CZE	8	178.0	50.250000
433	tomoaki takata	JPN	8	175.0	29.250000
304	nathan phillips	GBR	8	NaN	42.375000
255	manuel cornu	FRA	8	177.0	20.000000
177	jan hojer	GER	8	188.0	8.875000
140	gregor vezonik	SLO	8	176.0	38.875000
382	sergii topishko	UKR	8	NaN	47.375000
223	kevin heiniger	SUI	7	NaN	71.714286

In [63]:

df_m_boulder_agg.describe()

Out[63]:

	years in ranking	height	avg. rank
count	482.000000	127.000000	482.000000
mean	2.331950	174.574803	75.544836
std	1.984811	5.902004	36.109175
min	1.000000	162.000000	1.000000
25%	1.000000	170.000000	48.000000
50%	1.000000	175.000000	74.000000
75%	3.000000	178.000000	103.750000
max	10.000000	198.000000	149.000000

In [64]:

df_m_boulder_agg.loc[df_m_boulder_agg['years in ranking'] >= 5].describe()

Out[64]:

	years in ranking	height	avg. rank
count	67.000000	37.000000	67.000000
mean	6.537313	175.648649	40.344942
std	1.636037	5.735931	22.389098
min	5.000000	163.000000	7.666667
25%	5.000000	172.000000	23.250000
50%	6.000000	176.000000	38.285714
75%	8.000000	178.000000	53.600000
max	10.000000	188.000000	100.500000

In [65]:

df_m_boulder_agg.loc[df_m_boulder_agg['height'] > 185]

Out[65]:

	full_name	country	years in ranking	height	avg. rank
177	jan hojer	GER	8	188.0	8.875000
283	meichi narasaki	JPN	6	188.0	27.166667
2	adam ondra	CZE	5	186.0	10.400000
327	paul jenft	FRA	3	198.0	34.000000
75	christoph schweiger	GER	3	187.0	50.666667

In [66]:

df_m_boulder_agg.loc[df_m_boulder_agg['height'] < 165]

Out[66]:

	full_name	country	years in ranking	height	avg. rank
377	sean bailey	USA	6	163.0	22.333333
375	sascha lehmann	SUI	5	164.0	51.000000
388	shion omata	JPN	1	162.0	113.000000
354	ritsu kayotani	JPN	1	163.0	12.000000
96	dillon countryman	USA	1	164.0	41.000000

In [67]:

mean = df_m_boulder_agg['height'].mean()
std = df_m_boulder_agg['height'].std()
bins = np.arange(np.floor(df_m_boulder_agg['height'].min()), df_m_boulder_agg['height'].max()+1, 1)

plt.hist(df_m_boulder_agg['height'], bins=bins, color='deepskyblue', zorder=10, alpha=0.9)
plt.axvline(mean, color='black', lw=0.9)
plt.axvline(mean-std, linestyle='--', color='black', lw=0.9)
plt.axvline(mean+std, linestyle='--', color='black', lw=0.9)
ax = plt.gca()
ax.set_xticks(bins[::3])
plt.text(0.61, 0.93, f'Avg. height is {round(mean, 1)} $\pm$ {round(std, 1)}', transform=ax.transAxes)
plt.title('Height distribution of male climbers registered in the IFSC\n'+
          'Bouldering World Cup ranking between 2013 and 2023')
plt.xlabel('height [cm]')
plt.ylabel('number of climbers')
plt.show()

In [68]:

df_m_boulder_agg_clean = df_m_boulder_agg.loc[(df_m_boulder_agg['height'].notnull())
                                              & (df_m_boulder_agg['height'] < 195)]
stats.normaltest(df_m_boulder_agg_clean['height'])

Out[68]:

NormaltestResult(statistic=0.6821195834831839, pvalue=0.7110163940269245)

In [69]:

stats.ks_1samp(df_m_boulder_agg_clean['height'], stats.norm.cdf, args=(world_m_height_avg, world_m_height_std))

Out[69]:

KstestResult(statistic=0.29879267568448625, pvalue=1.8028202950648584e-10, statistic_location=178.0, statistic_sign=1)

In [70]:

stats.ttest_1samp(df_m_boulder_agg_clean['height'], world_m_height_avg)

Out[70]:

TtestResult(statistic=-8.127721561300902, pvalue=3.6372076149100477e-13, df=125)

In [71]:

df_m_boulder_agg_country = df_m_boulder_agg.groupby(['country'], as_index=False)
df_m_boulder_agg_country = df_m_boulder_agg_country.agg({'full_name': 'count', 'height': 'mean'})
df_m_boulder_agg_country = df_m_boulder_agg_country.sort_values('full_name', ascending=False)
df_m_boulder_agg_country.rename(columns={'full_name': 'no. of climbers', 'height': 'avg. height'})

Out[71]:

	country	no. of climbers	avg. height
54	USA	47	174.666667
27	JPN	44	171.555556
15	FRA	27	177.750000
18	GER	26	177.250000
7	CAN	26	172.750000
16	GBR	23	175.166667
2	AUT	20	178.000000
29	KOR	18	172.000000
47	SUI	16	171.800000
44	RUS	13	174.500000
9	CHN	12	173.000000
26	ITA	12	172.600000
23	IRI	12	173.000000
46	SLO	11	177.250000
13	ESP	11	178.250000
35	MEX	11	NaN
1	AUS	10	165.000000
25	ISR	9	171.250000
22	IND	9	NaN
37	NED	8	182.000000
4	BEL	7	174.000000
45	SGP	7	NaN
53	UKR	7	179.000000
8	CHI	6	180.000000
10	CZE	6	178.333333
41	POL	6	175.000000
19	HKG	6	172.000000
49	SWE	5	177.000000
14	FIN	5	180.000000
3	AZE	4	NaN
50	THA	4	NaN
31	LTU	4	176.000000
24	IRL	4	173.000000
5	BRA	4	NaN
17	GEO	4	NaN
52	TUR	3	NaN
39	NOR	3	175.000000
38	NEP	3	NaN
28	KAZ	3	170.000000
34	MAS	3	NaN
30	LAT	3	181.000000
20	HUN	3	176.000000
11	DEN	3	NaN
43	RSA	2	178.000000
6	BUL	2	172.000000
36	MRI	1	176.000000
40	PER	1	NaN
42	PUR	1	NaN
33	MAC	1	NaN
32	LUX	1	NaN
21	INA	1	NaN
48	SVK	1	175.000000
12	ECU	1	175.000000
51	TPE	1	NaN
0	ARG	1	NaN

Changes in the statistics over the years¶

In [72]:

years = list(range(2013, 2024))
years.pop(years.index(2020))
b_height_list = []
b_height_std_list = []

for year in years:
    heights = df_m_boulder.loc[df_m_boulder['year'] == year]['height']
    b_height_list.append(heights.mean())
    b_height_std_list.append(heights.std())
    
    avg = round(b_height_list[-1], 1)
    std = round(b_height_std_list[-1], 1)
    all_ = len(heights)
    nans = len(heights.loc[heights.isnull()])
    perc = round(100.0*nans/all_, 1)
    print(f'Avg. height in {year}: {avg} +- {std} '+
          f'({all_-nans} records, {nans} NaN, {perc}% of NaNs in {all_} climbers)')

Avg. height in 2013: 176.3 +- 5.1 (13 records, 102 NaN, 88.7% of NaNs in 115 climbers)
Avg. height in 2014: 175.4 +- 5.6 (21 records, 118 NaN, 84.9% of NaNs in 139 climbers)
Avg. height in 2015: 176.0 +- 5.3 (21 records, 61 NaN, 74.4% of NaNs in 82 climbers)
Avg. height in 2016: 176.1 +- 5.3 (28 records, 91 NaN, 76.5% of NaNs in 119 climbers)
Avg. height in 2017: 175.5 +- 5.4 (34 records, 77 NaN, 69.4% of NaNs in 111 climbers)
Avg. height in 2018: 175.4 +- 5.5 (37 records, 53 NaN, 58.9% of NaNs in 90 climbers)
Avg. height in 2019: 175.8 +- 5.8 (46 records, 39 NaN, 45.9% of NaNs in 85 climbers)
Avg. height in 2021: 175.3 +- 6.5 (54 records, 17 NaN, 23.9% of NaNs in 71 climbers)
Avg. height in 2022: 175.1 +- 5.8 (83 records, 70 NaN, 45.8% of NaNs in 153 climbers)
Avg. height in 2023: 174.7 +- 6.1 (94 records, 65 NaN, 40.9% of NaNs in 159 climbers)

In [73]:

for year in years:
    df_y = df_m_boulder.loc[df_m_boulder['year'] == year]
    avg_rank = round(df_y['rank'].mean())
    avg_rank_nan = round(df_y.loc[df_y['height'].isnull()]['rank'].mean())
    avg_rank_not_nan = round(df_y.loc[df_y['height'].notnull()]['rank'].mean())
    print(f'{year} avg. rank: {avg_rank}; avg. rank of climbers w/out height {avg_rank_nan},'+
          f' and with height data {avg_rank_not_nan}')

2013 avg. rank: 58; avg. rank of climbers w/out height 60, and with height data 39
2014 avg. rank: 70; avg. rank of climbers w/out height 75, and with height data 38
2015 avg. rank: 41; avg. rank of climbers w/out height 48, and with height data 22
2016 avg. rank: 59; avg. rank of climbers w/out height 66, and with height data 39
2017 avg. rank: 55; avg. rank of climbers w/out height 63, and with height data 38
2018 avg. rank: 45; avg. rank of climbers w/out height 53, and with height data 33
2019 avg. rank: 43; avg. rank of climbers w/out height 51, and with height data 36
2021 avg. rank: 36; avg. rank of climbers w/out height 53, and with height data 30
2022 avg. rank: 77; avg. rank of climbers w/out height 97, and with height data 59
2023 avg. rank: 79; avg. rank of climbers w/out height 108, and with height data 60

In [74]:

outlier_marker = dict(markerfacecolor='none', marker='o', markeredgecolor='orangered', alpha=0.7)
meanprops = dict(markerfacecolor='none', marker='x', markeredgecolor='gray', alpha=0.9, markersize=4)

df_m_boulder.boxplot(column='height', by='year', flierprops=outlier_marker, showmeans=True, meanprops=meanprops)
ax = plt.gca()
ax.grid(linestyle='--', alpha=0.5)
ax.grid(axis='x')
ax.set_yticks(np.arange(160, 200, 5))
plt.suptitle('')
plt.title('Box plot of height distribution of male climbers registered\n'+
          'in the IFSC Bouldering World Cup ranking for different years')
plt.xlabel('year')
plt.ylabel('height [cm]')
plt.show()

Differences between lead and boulder for men¶

In [75]:

mean = df_m_lead_agg['height'].mean()
std = df_m_lead_agg['height'].std()
bins = np.arange(np.floor(df_m_lead_agg['height'].min()), df_m_lead_agg['height'].max()+1, 1)
plt.hist(df_m_lead_agg['height'], bins=bins, color='green', zorder=10, alpha=0.5, label='lead')
plt.axvline(mean, color='black', lw=0.95, linestyle='-.')

mean_b = df_m_boulder_agg['height'].mean()
std_b = df_m_boulder_agg['height'].std()
bins_b = np.arange(np.floor(df_m_boulder_agg['height'].min()), df_m_boulder_agg['height'].max()+1, 1)
plt.hist(df_m_boulder_agg['height'], bins=bins_b, color='darkorchid', zorder=10, alpha=0.5, label='boulder')
plt.axvline(mean_b, color='black', lw=0.95, linestyle='--')

plt.legend(loc=2)
ax = plt.gca()
ax.set_xticks(bins[::3])
plt.text(0.64, 0.82, f'Avg. height is:\n{round(mean, 1)} $\pm$ {round(std, 1)} in lead\n'+
         f'{round(mean_b, 1)} $\pm$ {round(std_b, 1)} in boulder', transform=ax.transAxes)
plt.title('Height distribution of male climbers registered in\nthe IFSC World Cup rankings between 2013 and 2023')
plt.xlabel('height [cm]')
plt.ylabel('number of climbers')
plt.show()

In [76]:

b_height_list = np.array(b_height_list)
b_height_std_list = np.array(b_height_std_list)

plt.plot(years, b_height_list, color='darkorchid', lw=1.8, label='boulder')
plt.fill_between(years, b_height_list - b_height_std_list, b_height_list + b_height_std_list,
                 color='darkorchid', lw=0, alpha=0.2)

l_height_list = np.array(l_height_list)
l_height_std_list = np.array(l_height_std_list)

plt.plot(years, l_height_list, color='green', lw=1.8, label='lead')
plt.fill_between(years, l_height_list - l_height_std_list, l_height_list + l_height_std_list,
                 color='green', lw=0, alpha=0.2)

plt.xlim([min(years), max(years)])
plt.ylim([164, 185.5])
plt.legend()
ax = plt.gca()
ax.set_xticks(years)
ax.set_yticks(np.arange(165, 186, 2))
plt.title('Comparison of the avg. height of male climbers\nin lead and bouldering IFSC rankings over the years')
plt.xlabel('year')
plt.ylabel('height [cm]')
plt.show()

Jump in height from 2018 to 2019 for lead¶

In [77]:

df_m_lead_18_19 = df_m_lead.loc[(df_m_lead['year'].isin([2018, 2019])) & (df_m_lead['height'].notnull())]
df_m_lead_18_19 = df_m_lead_18_19.groupby(['full_name'], as_index=False)
df_m_lead_18_19 = df_m_lead_18_19.agg({'full_name': 'first', 'country': 'first',
                                       'year': 'sum', 'height': 'mean'})
df_m_lead_18 = df_m_lead_18_19.loc[df_m_lead_18_19['year'] == 2018]
df_m_lead_19 = df_m_lead_18_19.loc[df_m_lead_18_19['year'] == 2019]

print(f'Avg. height of climbers competeing in 2018 but not in 2019: {df_m_lead_18["height"].mean()}')
df_m_lead_18.sort_values('height')

Avg. height of climbers competeing in 2018 but not in 2019: 172.45454545454547

Out[77]:

	full_name	country	year	height
17	hyunbin min	KOR	2018	162.0
61	veddriq leonardo	INA	2018	162.0
32	marcin dzienski	POL	2018	166.0
21	jeremy bonder	FRA	2018	168.0
35	masahiro higuchi	JPN	2018	169.0
63	yoshiyuki ogata	JPN	2018	172.0
37	max kleesattel	GER	2018	173.0
49	romaric geffroy	FRA	2018	177.0
24	john brosler	USA	2018	179.0
6	arsène duval	FRA	2018	181.0
29	louis gundolf	AUT	2018	188.0

In [78]:

print(f'Avg. height of climbers competeing in 2019 but not in 2018: {df_m_lead_19["height"].mean()}')
df_m_lead_19.sort_values('height')

Avg. height of climbers competeing in 2019 but not in 2018: 177.35

Out[78]:

	full_name	country	year	height
45	nimrod marcus	ISR	2019	166.0
66	zach galla	USA	2019	168.0
57	sungsu lee	KOR	2019	170.0
10	dmitrii fakirianov	RUS	2019	171.0
48	rei sugimoto	JPN	2019	172.0
64	yufei pan	CHN	2019	173.0
47	philipp martin	GER	2019	173.0
11	dohyun lee	KOR	2019	174.0
60	tomoaki takata	JPN	2019	175.0
5	anze peharc	SLO	2019	177.0
25	jongwon chon	KOR	2019	177.0
34	martin stranik	CZE	2019	178.0
40	mickael mawem	FRA	2019	179.0
23	jesse grupper	USA	2019	180.0
4	alistair duval	FRA	2019	180.0
7	campbell harrison	AUS	2019	182.0
43	nathaniel coleman	USA	2019	182.0
41	mikel asier linacisoro molina	ESP	2019	184.0
38	meichi narasaki	JPN	2019	188.0
46	paul jenft	FRA	2019	198.0

Jump in height from 2018 to 2019 for bouldering¶

In [79]:

df_m_boulder_18_19 = df_m_boulder.loc[(df_m_boulder['year'].isin([2018, 2019])) & (df_m_boulder['height'].notnull())]
df_m_boulder_18_19 = df_m_boulder_18_19.groupby(['full_name'], as_index=False)
df_m_boulder_18_19 = df_m_boulder_18_19.agg({'full_name': 'first', 'country': 'first',
                                             'year': 'sum', 'height': 'mean'})
df_m_boulder_18 = df_m_boulder_18_19.loc[df_m_boulder_18_19['year'] == 2018]
df_m_boulder_19 = df_m_boulder_18_19.loc[df_m_boulder_18_19['year'] == 2019]

print(f'Avg. height of climbers competeing in 2018 but not in 2019: {df_m_boulder_18["height"].mean()}')
df_m_boulder_18.sort_values('height')

Avg. height of climbers competeing in 2018 but not in 2019: 173.0

Out[79]:

	full_name	country	year	height
32	nimrod marcus	ISR	2018	166.0
22	masahiro higuchi	JPN	2018	169.0
42	thilo jeldrik schröter	NOR	2018	175.0
31	nils favre	SUI	2018	176.0
49	zan lovenjak sudar	SLO	2018	179.0

In [80]:

print(f'Avg. height of climbers competeing in 2019 but not in 2018: {df_m_boulder_19["height"].mean()}')
df_m_boulder_19.sort_values('height')

Avg. height of climbers competeing in 2019 but not in 2018: 175.92857142857142

Out[80]:

	full_name	country	year	height
40	simon lorenzi	BEL	2019	168.0
48	zach galla	USA	2019	168.0
41	stefano ghisolfi	ITA	2019	169.0
34	ram levin	ISR	2019	171.0
35	rei kawamata	JPN	2019	172.0
29	nicolai uznik	AUT	2019	173.0
33	philipp martin	GER	2019	173.0
5	carlos felipe granja lopez	ECU	2019	175.0
18	luka potocar	SLO	2019	177.0
1	alberto ginés lópez	ESP	2019	178.0
50	zander waller	USA	2019	182.0
27	mikel asier linacisoro molina	ESP	2019	184.0
0	adam ondra	CZE	2019	186.0
7	christoph schweiger	GER	2019	187.0

Basic statistics for both categories for women¶

In [81]:

df_cup_f = df_cup.loc[df_cup['sex'] == 'F']
df_cup_f.info()

<class 'pandas.core.frame.DataFrame'>
Index: 2172 entries, 313 to 4357
Data columns (total 7 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   year       2172 non-null   int64  
 1   category   2172 non-null   object 
 2   sex        2172 non-null   object 
 3   full_name  2172 non-null   object 
 4   height     928 non-null    float64
 5   country    2172 non-null   object 
 6   rank       2172 non-null   int64  
dtypes: float64(1), int64(2), object(4)
memory usage: 135.8+ KB

In [82]:

column = df_cup_f.copy()['category']
df_cup_f.insert(0, 'category_dup', column, allow_duplicates=False)
df_cup_f_agg = df_cup_f.groupby(['full_name'], as_index=False)
df_cup_f_agg = df_cup_f_agg.agg({'full_name': 'first', 'country': 'first', 'year': 'count',
                                 'category': lambda x: (x == 'lead').sum(),
                                 'category_dup': lambda x: (x == 'boulder').sum(),
                                 'height': 'mean', 'rank': 'mean'})
df_cup_f_agg = df_cup_f_agg.sort_values('year', ascending=False)
df_cup_f_agg = df_cup_f_agg.rename(columns={'year': 'times in ranking', 'rank' : 'avg. rank',
                                            'category': 'in lead', 'category_dup': 'in boulder'})

df_cup_f_agg.info()

<class 'pandas.core.frame.DataFrame'>
Index: 687 entries, 116 to 513
Data columns (total 7 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   full_name         687 non-null    object 
 1   country           687 non-null    object 
 2   times in ranking  687 non-null    int64  
 3   in lead           687 non-null    int64  
 4   in boulder        687 non-null    int64  
 5   height            159 non-null    float64
 6   avg. rank         687 non-null    float64
dtypes: float64(2), int64(3), object(2)
memory usage: 42.9+ KB

In [83]:

df_cup_f_agg.head(20)

Out[83]:

	full_name	country	times in ranking	in lead	in boulder	height	avg. rank
116	chloe caulier	BEL	17	7	10	166.0	49.764706
314	katja debevec	SLO	17	8	9	173.0	37.705882
280	jessica pilz	AUT	17	10	7	165.0	14.529412
441	miho nonaka	JPN	16	7	9	163.0	16.687500
8	akiyo noguchi	JPN	16	8	8	167.0	8.687500
189	fanny gibert	FRA	15	5	10	165.0	20.666667
291	julia chanourdie	FRA	15	10	5	164.0	34.733333
271	janja garnbret	SLO	15	8	7	164.0	4.533333
35	andrea kümin	SUI	15	7	8	164.0	57.400000
590	sol sa	KOR	15	6	9	NaN	32.333333
460	molly thompson-smith	GBR	14	10	4	159.0	42.714286
328	kyra condie	USA	14	5	9	162.0	41.642857
518	petra klingler	SUI	14	4	10	162.0	26.000000
59	anne-sophie koller	SUI	14	10	4	160.0	55.214286
247	ievgeniia kazbekova	UKR	14	8	6	164.0	30.928571
267	jain kim	KOR	13	8	5	152.0	16.000000
650	vita lukan	SLO	12	7	5	164.0	24.333333
598	stasa gejo	SRB	12	5	7	175.0	28.250000
95	brooke raboutou	USA	12	6	6	158.0	26.333333
432	mia krampl	SLO	12	7	5	163.0	25.083333

In [84]:

df_cup_f_agg.describe()

Out[84]:

	times in ranking	in lead	in boulder	height	avg. rank
count	687.000000	687.000000	687.000000	159.000000	687.000000
mean	3.161572	1.500728	1.660844	163.270440	73.178412
std	3.124031	2.002094	1.921179	5.880331	32.171613
min	1.000000	0.000000	0.000000	149.000000	4.533333
25%	1.000000	0.000000	0.000000	160.000000	51.298611
50%	2.000000	1.000000	1.000000	163.000000	72.000000
75%	4.000000	2.000000	2.000000	167.500000	94.000000
max	17.000000	10.000000	10.000000	181.000000	158.000000

In [85]:

df_cup_f_agg_country = df_cup_f_agg.groupby(['country'], as_index=False)
df_cup_f_agg_country = df_cup_f_agg_country.agg({'full_name': 'count', 'height': 'mean'})
df_cup_f_agg_country = df_cup_f_agg_country.sort_values('full_name', ascending=False)
df_cup_f_agg_country.rename(columns={'full_name': 'no. of climbers', 'height': 'avg. height'})
df_cup_f_agg_country.head(30)

Out[85]:

	country	full_name	height
61	USA	67	164.857143
33	JPN	58	160.055556
18	FRA	48	162.222222
8	CAN	33	165.000000
21	GER	31	164.777778
35	KOR	31	160.625000
50	RUS	30	163.250000
2	AUT	30	165.900000
32	ITA	29	162.571429
19	GBR	27	165.000000
52	SLO	25	164.307692
10	CHN	18	161.333333
54	SUI	18	161.833333
27	INA	17	155.500000
12	CZE	16	162.000000
43	NOR	16	164.000000
1	AUS	14	171.333333
28	IND	13	NaN
47	POL	12	168.000000
29	IRI	11	162.000000
4	BEL	10	162.500000
60	UKR	9	159.000000
16	ESP	9	NaN
51	SGP	8	169.000000
31	ISR	8	166.500000
38	MEX	7	NaN
56	SWE	7	NaN
17	FIN	6	NaN
5	BRA	6	164.000000
41	NED	6	165.500000

Analysis of the lead world cup data for women¶

In [86]:

df_f_lead = df_cup.loc[(df_cup['sex'] == 'F') & (df_cup['category'] == 'lead')]
df_f_lead.info()

<class 'pandas.core.frame.DataFrame'>
Index: 1031 entries, 313 to 4239
Data columns (total 7 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   year       1031 non-null   int64  
 1   category   1031 non-null   object 
 2   sex        1031 non-null   object 
 3   full_name  1031 non-null   object 
 4   height     489 non-null    float64
 5   country    1031 non-null   object 
 6   rank       1031 non-null   int64  
dtypes: float64(1), int64(2), object(4)
memory usage: 64.4+ KB

In [87]:

df_f_lead.describe()
# note that in the numbers below a climber who was ranked for several years
# is taken into account that many times in the calculations

Out[87]:

	year	height	rank
count	1031.000000	489.000000	1031.000000
mean	2018.260912	162.188139	54.704171
std	3.422203	5.565683	34.858978
min	2013.000000	149.000000	1.000000
25%	2015.000000	159.000000	26.000000
50%	2018.000000	162.000000	52.000000
75%	2022.000000	165.000000	77.000000
max	2023.000000	176.000000	150.000000

Analysis of aggregated lead rankings for all years¶

In [88]:

df_f_lead_agg = df_f_lead.groupby(['full_name'], as_index=False)
df_f_lead_agg = df_f_lead_agg.agg({'full_name': 'first', 'country': 'first', 'year': 'count',
                                   'height': 'mean', 'rank': 'mean'})
df_f_lead_agg = df_f_lead_agg.sort_values('year', ascending=False)
df_f_lead_agg = df_f_lead_agg.rename(columns={'year': 'years in ranking', 'rank' : 'avg. rank'})

df_f_lead_agg.info()

<class 'pandas.core.frame.DataFrame'>
Index: 427 entries, 349 to 426
Data columns (total 5 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   full_name         427 non-null    object 
 1   country           427 non-null    object 
 2   years in ranking  427 non-null    int64  
 3   height            130 non-null    float64
 4   avg. rank         427 non-null    float64
dtypes: float64(2), int64(1), object(2)
memory usage: 20.0+ KB

In [89]:

df_f_lead_agg.head(20)

Out[89]:

	full_name	country	years in ranking	height	avg. rank
349	salomé romain	FRA	10	149.0	23.200000
171	jessica pilz	AUT	10	165.0	11.200000
179	julia chanourdie	FRA	10	164.0	23.600000
283	molly thompson-smith	GBR	10	159.0	30.300000
68	claudia ghisolfi	ITA	10	162.0	40.900000
32	anne-sophie koller	SUI	10	160.0	45.700000
181	julia fiser	AUT	9	158.0	51.222222
384	tina johnsen hafsaas	NOR	9	164.0	43.333333
143	hélène janicot	FRA	8	165.0	11.875000
270	michelle hulliger	SUI	8	157.0	50.500000
161	jain kim	KOR	8	152.0	3.250000
146	ievgeniia kazbekova	UKR	8	164.0	30.750000
339	risa ota	JPN	8	161.0	27.750000
192	katja debevec	SLO	8	173.0	46.625000
164	janja garnbret	SLO	8	164.0	2.000000
6	akiyo noguchi	JPN	8	167.0	14.375000
401	vita lukan	SLO	7	164.0	15.714286
234	magdalena röck	AUT	7	NaN	30.428571
17	anak verhoeven	BEL	7	NaN	8.428571
275	mina markovic	SLO	7	161.0	7.285714

In [90]:

df_f_lead_agg.describe()

Out[90]:

	years in ranking	height	avg. rank
count	427.000000	130.000000	427.000000
mean	2.414520	162.869231	68.605569
std	2.059918	5.542491	32.570639
min	1.000000	149.000000	2.000000
25%	1.000000	160.000000	45.850000
50%	2.000000	163.000000	68.000000
75%	3.000000	167.000000	87.500000
max	10.000000	176.000000	150.000000

In [91]:

df_f_lead_agg.loc[df_f_lead_agg['years in ranking'] >= 5].describe()

Out[91]:

	years in ranking	height	avg. rank
count	64.000000	42.000000	64.000000
mean	6.656250	162.000000	36.275961
std	1.545282	5.512735	18.338856
min	5.000000	149.000000	2.000000
25%	5.000000	159.000000	23.500000
50%	6.000000	162.500000	34.500000
75%	7.250000	165.000000	50.000000
max	10.000000	175.000000	75.666667

In [92]:

df_f_lead_agg.loc[df_f_lead_agg['height'] > 172]

Out[92]:

	full_name	country	years in ranking	height	avg. rank
192	katja debevec	SLO	8	173.0	46.625
371	stasa gejo	SRB	5	175.0	52.800
314	oceania mackenzie	AUS	4	173.0	58.250
180	julia duffy	USA	2	176.0	60.000
184	julija kruder	SLO	2	175.0	96.000
108	flavy cohaut	FRA	1	174.0	128.000

In [93]:

df_f_lead_agg.loc[df_f_lead_agg['height'] < 153]

Out[93]:

	full_name	country	years in ranking	height	avg. rank
349	salomé romain	FRA	10	149.0	23.200000
161	jain kim	KOR	8	152.0	3.250000
74	dinara fakhritdinova	RUS	7	152.0	17.571429
206	laura rogora	ITA	6	152.0	15.166667
336	rebeka kamin	SLO	4	152.0	42.250000
297	natsuki tanii	JPN	4	152.0	5.750000
280	miu kakizaki	JPN	3	149.0	49.000000

In [94]:

mean = df_f_lead_agg['height'].mean()
std = df_f_lead_agg['height'].std()
bins = np.arange(np.floor(df_f_lead_agg['height'].min()), df_f_lead_agg['height'].max()+1, 1)

plt.hist(df_f_lead_agg['height'], bins=bins, color='tomato', zorder=10, alpha=0.9)
plt.axvline(mean, color='black', lw=0.9)
plt.axvline(mean-std, linestyle='--', color='black', lw=0.9)
plt.axvline(mean+std, linestyle='--', color='black', lw=0.9)
ax = plt.gca()
ax.set_xticks(bins[::3])
plt.text(0.05, 0.87, f'Avg. height\nis {round(mean, 1)} $\pm$ {round(std, 1)}', transform=ax.transAxes)
plt.title('Height distribution of female climbers registered in\nthe IFSC '+
          'Lead World Cup ranking between 2013 and 2023')
plt.xlabel('height [cm]')
plt.ylabel('number of climbers')
plt.show()

In [95]:

df_f_lead_agg_clean = df_f_lead_agg.loc[(df_f_lead_agg['height'].notnull())]
stats.normaltest(df_f_lead_agg_clean['height'])

Out[95]:

NormaltestResult(statistic=0.2331952749381621, pvalue=0.8899432009328268)

In [96]:

stats.ks_1samp(df_f_lead_agg_clean['height'], stats.norm.cdf, args=(world_f_height_avg, world_f_height_std))

Out[96]:

KstestResult(statistic=0.20615375784824586, pvalue=2.533600800064976e-05, statistic_location=165.0, statistic_sign=1)

In [97]:

stats.ttest_1samp(df_f_lead_agg_clean['height'], world_f_height_avg)

Out[97]:

TtestResult(statistic=-3.766173501894482, pvalue=0.0002508890927849896, df=129)

In [98]:

df_f_lead_agg_country = df_f_lead_agg.groupby(['country'], as_index=False)
df_f_lead_agg_country = df_f_lead_agg_country.agg({'full_name': 'count', 'height': 'mean'})
df_f_lead_agg_country = df_f_lead_agg_country.sort_values('full_name', ascending=False)
df_f_lead_agg_country.rename(columns={'full_name': 'no. of climbers', 'height': 'avg. height'})

Out[98]:

	country	no. of climbers	avg. height
25	JPN	36	159.800000
14	FRA	31	162.357143
47	USA	31	164.300000
27	KOR	25	161.200000
39	SLO	24	164.307692
16	GER	23	164.000000
37	RUS	22	160.666667
15	GBR	19	165.000000
24	ITA	19	159.200000
8	CHN	17	161.333333
2	AUT	17	164.625000
19	INA	16	155.500000
41	SUI	15	161.833333
9	CZE	12	162.000000
6	CAN	11	163.000000
31	NOR	11	164.000000
3	BEL	8	162.500000
34	POL	7	168.000000
1	AUS	7	166.500000
23	ISR	6	166.500000
7	CHI	5	160.666667
12	ESP	5	NaN
4	BRA	5	164.000000
30	NED	4	165.500000
43	SWE	4	NaN
17	HKG	4	NaN
44	THA	4	NaN
46	UKR	3	159.000000
32	NZL	3	NaN
0	ARG	3	157.000000
20	IND	3	NaN
28	MEX	3	NaN
42	SVK	2	169.000000
45	TPE	2	167.000000
11	ECU	2	NaN
36	RSA	2	NaN
21	IRI	2	161.000000
38	SGP	2	NaN
40	SRB	2	175.000000
5	BUL	1	162.000000
10	DEN	1	NaN
13	FIN	1	NaN
29	MKD	1	NaN
18	HUN	1	NaN
35	PUR	1	NaN
22	ISL	1	171.000000
33	PER	1	168.000000
26	KAZ	1	NaN
48	VEN	1	NaN

Changes in the statistics over the years¶

In [99]:

years = list(range(2013, 2024))
years.pop(years.index(2020))
lf_height_list = []
lf_height_std_list = []

for year in years:
    heights = df_f_lead.loc[df_f_lead['year'] == year]['height']
    lf_height_list.append(heights.mean())
    lf_height_std_list.append(heights.std())
    
    avg = round(lf_height_list[-1], 1)
    std = round(lf_height_std_list[-1], 1)
    all_ = len(heights)
    nans = len(heights.loc[heights.isnull()])
    perc = round(100.0*nans/all_, 1)
    print(f'Avg. height in {year}: {avg} +- {std} '+
          f'({all_-nans} records, {nans} NaN, {perc}% of NaNs in {all_} climbers)')

Avg. height in 2013: 161.3 +- 5.9 (19 records, 77 NaN, 80.2% of NaNs in 96 climbers)
Avg. height in 2014: 160.9 +- 6.7 (21 records, 78 NaN, 78.8% of NaNs in 99 climbers)
Avg. height in 2015: 161.0 +- 6.1 (22 records, 62 NaN, 73.8% of NaNs in 84 climbers)
Avg. height in 2016: 161.2 +- 5.6 (33 records, 49 NaN, 59.8% of NaNs in 82 climbers)
Avg. height in 2017: 162.2 +- 5.6 (44 records, 64 NaN, 59.3% of NaNs in 108 climbers)
Avg. height in 2018: 162.7 +- 5.8 (49 records, 51 NaN, 51.0% of NaNs in 100 climbers)
Avg. height in 2019: 161.8 +- 5.0 (55 records, 30 NaN, 35.3% of NaNs in 85 climbers)
Avg. height in 2021: 162.3 +- 5.4 (67 records, 11 NaN, 14.1% of NaNs in 78 climbers)
Avg. height in 2022: 162.7 +- 5.6 (86 records, 59 NaN, 40.7% of NaNs in 145 climbers)
Avg. height in 2023: 162.7 +- 5.3 (93 records, 61 NaN, 39.6% of NaNs in 154 climbers)

In [100]:

for year in years:
    df_y = df_f_lead.loc[df_f_lead['year'] == year]
    avg_rank = round(df_y['rank'].mean())
    avg_rank_nan = round(df_y.loc[df_y['height'].isnull()]['rank'].mean())
    avg_rank_not_nan = round(df_y.loc[df_y['height'].notnull()]['rank'].mean())
    print(f'{year} avg. rank: {avg_rank}; avg. rank of climbers w/out height {avg_rank_nan},'+
          f' and with height data {avg_rank_not_nan}')

2013 avg. rank: 48; avg. rank of climbers w/out height 51, and with height data 35
2014 avg. rank: 50; avg. rank of climbers w/out height 53, and with height data 39
2015 avg. rank: 42; avg. rank of climbers w/out height 47, and with height data 28
2016 avg. rank: 41; avg. rank of climbers w/out height 43, and with height data 39
2017 avg. rank: 54; avg. rank of climbers w/out height 66, and with height data 38
2018 avg. rank: 50; avg. rank of climbers w/out height 58, and with height data 41
2019 avg. rank: 43; avg. rank of climbers w/out height 55, and with height data 36
2021 avg. rank: 39; avg. rank of climbers w/out height 63, and with height data 35
2022 avg. rank: 73; avg. rank of climbers w/out height 105, and with height data 50
2023 avg. rank: 77; avg. rank of climbers w/out height 107, and with height data 57

In [101]:

outlier_marker = dict(markerfacecolor='none', marker='o', markeredgecolor='orangered', alpha=0.7)
meanprops = dict(markerfacecolor='none', marker='x', markeredgecolor='gray', alpha=0.9, markersize=4)
boxprops=dict(color='darkslateblue')
whiskerprops=boxprops

df_f_lead.boxplot(column='height', by='year', flierprops=outlier_marker, showmeans=True,
                  meanprops=meanprops, boxprops=boxprops, whiskerprops=whiskerprops)
ax = plt.gca()
ax.grid(linestyle='--', alpha=0.5)
ax.grid(axis='x')
ax.set_yticks(np.arange(149, 178, 3))
plt.suptitle('')
plt.title('Box plot of height distribution of female climbers registered\n'+
          'in the IFSC Lead World Cup ranking for different years')
plt.xlabel('year')
plt.ylabel('height [cm]')
plt.show()

Analysis of the bouldring world cup data for women¶

In [102]:

df_f_boulder = df_cup.loc[(df_cup['sex'] == 'F') & (df_cup['category'] == 'boulder')]
df_f_boulder.info()

<class 'pandas.core.frame.DataFrame'>
Index: 1141 entries, 458 to 4357
Data columns (total 7 columns):
 #   Column     Non-Null Count  Dtype  
---  ------     --------------  -----  
 0   year       1141 non-null   int64  
 1   category   1141 non-null   object 
 2   sex        1141 non-null   object 
 3   full_name  1141 non-null   object 
 4   height     439 non-null    float64
 5   country    1141 non-null   object 
 6   rank       1141 non-null   int64  
dtypes: float64(1), int64(2), object(4)
memory usage: 71.3+ KB

In [103]:

df_f_boulder.describe()
# note that in the numbers below a climber who was ranked for several years
# is taken into account that many times in the calculations

Out[103]:

	year	height	rank
count	1141.000000	439.000000	1141.000000
mean	2017.991236	164.576310	60.514461
std	3.453439	5.567599	37.891269
min	2013.000000	150.000000	1.000000
25%	2015.000000	161.000000	29.000000
50%	2018.000000	164.000000	58.000000
75%	2022.000000	168.000000	88.000000
max	2023.000000	181.000000	158.000000

Analysis of aggregated bouldering rankings for all years¶

In [104]:

df_f_boulder_agg = df_f_boulder.groupby(['full_name'], as_index=False)
df_f_boulder_agg = df_f_boulder_agg.agg({'full_name': 'first', 'country': 'first', 'year': 'count',
                                   'height': 'mean', 'rank': 'mean'})
df_f_boulder_agg = df_f_boulder_agg.sort_values('year', ascending=False)
df_f_boulder_agg = df_f_boulder_agg.rename(columns={'year': 'years in ranking', 'rank' : 'avg. rank'})

df_f_boulder_agg.info()

<class 'pandas.core.frame.DataFrame'>
Index: 487 entries, 148 to 243
Data columns (total 5 columns):
 #   Column            Non-Null Count  Dtype  
---  ------            --------------  -----  
 0   full_name         487 non-null    object 
 1   country           487 non-null    object 
 2   years in ranking  487 non-null    int64  
 3   height            133 non-null    float64
 4   avg. rank         487 non-null    float64
dtypes: float64(2), int64(1), object(2)
memory usage: 22.8+ KB

In [105]:

df_f_boulder_agg.head(20)

Out[105]:

	full_name	country	years in ranking	height	avg. rank
148	fanny gibert	FRA	10	165.0	13.500000
366	petra klingler	SUI	10	162.0	13.800000
97	chloe caulier	BEL	10	166.0	33.400000
217	julija kruder	SLO	10	175.0	47.500000
420	sol sa	KOR	9	NaN	20.666667
319	miho nonaka	JPN	9	163.0	4.333333
237	kyra condie	USA	9	162.0	42.111111
228	katja debevec	SLO	9	173.0	29.777778
344	natalie bärtschi	SUI	9	NaN	72.000000
29	andrea kümin	SUI	8	164.0	56.250000
21	alma bestvater	GER	8	161.0	53.000000
408	shauna coxsey	GBR	8	164.0	11.375000
8	akiyo noguchi	JPN	8	167.0	3.000000
210	johanna färber	AUT	7	171.0	36.000000
158	franziska sterrer	AUT	7	169.0	30.857143
120	ekaterina kipriianova	RUS	7	NaN	39.000000
207	jessica pilz	AUT	7	165.0	19.285714
419	sofya yokoyama	SUI	7	NaN	59.714286
384	rong jiang	CHN	7	NaN	58.714286
184	hung ying lee	TPE	7	170.0	53.142857

In [106]:

df_f_boulder_agg.describe()

Out[106]:

	years in ranking	height	avg. rank
count	487.000000	133.000000	487.000000
mean	2.342916	163.947368	75.153275
std	1.899606	5.560025	35.488408
min	1.000000	150.000000	3.000000
25%	1.000000	160.000000	51.000000
50%	2.000000	164.000000	73.000000
75%	3.000000	168.000000	100.000000
max	10.000000	181.000000	158.000000

In [107]:

df_f_boulder_agg.loc[df_f_boulder_agg['years in ranking'] >= 5].describe()

Out[107]:

	years in ranking	height	avg. rank
count	66.000000	37.000000	66.000000
mean	6.378788	165.324324	39.960528
std	1.526838	5.701141	23.224164
min	5.000000	152.000000	3.000000
25%	5.000000	162.000000	23.850000
50%	6.000000	164.000000	36.300000
75%	7.000000	168.000000	55.500000
max	10.000000	181.000000	97.000000

In [108]:

df_f_boulder_agg.loc[df_f_boulder_agg['height'] > 172]

Out[108]:

	full_name	country	years in ranking	height	avg. rank
217	julija kruder	SLO	10	175.0	47.500000
228	katja debevec	SLO	9	173.0	29.777778
423	stasa gejo	SRB	7	175.0	10.714286
154	flavy cohaut	FRA	5	174.0	36.200000
137	emma horan	AUS	5	181.0	80.200000
357	oceania mackenzie	AUS	4	173.0	32.000000
411	sienna kopf	USA	3	174.0	47.666667

In [109]:

df_f_boulder_agg.loc[df_f_boulder_agg['height'] < 154]

Out[109]:

	full_name	country	years in ranking	height	avg. rank
199	jain kim	KOR	5	152.0	36.4
241	laura rogora	ITA	4	152.0	31.0
116	dinara fakhritdinova	RUS	2	152.0	22.5
346	natsuki tanii	JPN	1	152.0	51.0
90	chaeyeong kim	KOR	1	150.0	128.0

In [110]:

mean = df_f_boulder_agg['height'].mean()
std = df_f_boulder_agg['height'].std()
bins = np.arange(np.floor(df_f_boulder_agg['height'].min()), df_f_boulder_agg['height'].max()+1, 1)

plt.hist(df_f_boulder_agg['height'], bins=bins, color='tomato', zorder=10, alpha=0.9)
plt.axvline(mean, color='black', lw=0.9)
plt.axvline(mean-std, linestyle='--', color='black', lw=0.9)
plt.axvline(mean+std, linestyle='--', color='black', lw=0.9)
ax = plt.gca()
ax.set_xticks(bins[::3])
plt.text(0.05, 0.87, f'Avg. height\nis {round(mean, 1)} $\pm$ {round(std, 1)}', transform=ax.transAxes)
plt.title('Height distribution of female climbers registered in the IFSC\n'+
          'Bouldering World Cup ranking between 2013 and 2023')
plt.xlabel('height [cm]')
plt.ylabel('number of climbers')
plt.show()

In [111]:

df_f_boulder_agg_clean = df_f_boulder_agg.loc[(df_f_boulder_agg['height'].notnull())
                                              & (df_f_boulder_agg['height'] < 181)]
stats.normaltest(df_f_boulder_agg_clean['height'])

Out[111]:

NormaltestResult(statistic=1.2085361042148373, pvalue=0.5464742709802409)

In [112]:

stats.ks_1samp(df_f_boulder_agg_clean['height'], stats.norm.cdf, args=(world_f_height_avg, world_f_height_std))

Out[112]:

KstestResult(statistic=0.14216774386223185, pvalue=0.00862427465183941, statistic_location=165.0, statistic_sign=1)

In [113]:

stats.ttest_1samp(df_f_boulder_agg_clean['height'], world_f_height_avg)

Out[113]:

TtestResult(statistic=-1.8841588901953716, pvalue=0.06175961716324531, df=131)

In [114]:

df_f_boulder_agg_country = df_f_boulder_agg.groupby(['country'], as_index=False)
df_f_boulder_agg_country = df_f_boulder_agg_country.agg({'full_name': 'count', 'height': 'mean'})
df_f_boulder_agg_country = df_f_boulder_agg_country.sort_values('full_name', ascending=False)
df_f_boulder_agg_country.rename(columns={'full_name': 'no. of climbers', 'height': 'avg. height'})
df_f_boulder_agg_country.head(30)

Out[114]:

	country	full_name	height
61	USA	58	164.000000
33	JPN	44	161.000000
18	FRA	32	163.538462
8	CAN	29	164.333333
2	AUT	25	166.777778
21	GER	20	165.142857
32	ITA	19	163.000000
19	GBR	19	165.000000
50	RUS	17	163.250000
35	KOR	15	160.625000
52	SLO	13	166.600000
28	IND	11	NaN
29	IRI	11	162.000000
1	AUS	11	177.000000
10	CHN	11	160.000000
54	SUI	10	162.800000
43	NOR	9	164.000000
12	CZE	8	162.000000
38	MEX	7	NaN
47	POL	7	NaN
31	ISR	7	166.500000
51	SGP	7	169.000000
60	UKR	7	160.666667
17	FIN	6	NaN
16	ESP	6	NaN
41	NED	5	165.500000
57	THA	5	NaN
5	BRA	5	164.000000
55	SVK	4	169.000000
56	SWE	4	NaN

Changes in the statistics over the years¶

In [115]:

years = list(range(2013, 2024))
years.pop(years.index(2020))
bf_height_list = []
bf_height_std_list = []

for year in years:
    # Emma Horan increases the average by 1cm for 2014 and 2015, and 0.5cm for 2016-2018
    heights = df_f_boulder.loc[df_f_boulder['year'] == year]['height']
    bf_height_list.append(heights.mean())
    bf_height_std_list.append(heights.std())
    
    avg = round(bf_height_list[-1], 1)
    std = round(bf_height_std_list[-1], 1)
    all_ = len(heights)
    nans = len(heights.loc[heights.isnull()])
    perc = round(100.0*nans/all_, 1)
    print(f'Avg. height in {year}: {avg} +- {std} '+
          f'({all_-nans} records, {nans} NaN, {perc}% of NaNs in {all_} climbers)')

Avg. height in 2013: 162.6 +- 5.3 (17 records, 101 NaN, 85.6% of NaNs in 118 climbers)
Avg. height in 2014: 164.3 +- 7.3 (18 records, 112 NaN, 86.2% of NaNs in 130 climbers)
Avg. height in 2015: 167.6 +- 5.8 (16 records, 69 NaN, 81.2% of NaNs in 85 climbers)
Avg. height in 2016: 165.6 +- 5.8 (30 records, 92 NaN, 75.4% of NaNs in 122 climbers)
Avg. height in 2017: 165.6 +- 6.0 (36 records, 77 NaN, 68.1% of NaNs in 113 climbers)
Avg. height in 2018: 165.6 +- 5.7 (50 records, 53 NaN, 51.5% of NaNs in 103 climbers)
Avg. height in 2019: 163.7 +- 5.9 (55 records, 33 NaN, 37.5% of NaNs in 88 climbers)
Avg. height in 2021: 164.6 +- 5.1 (53 records, 15 NaN, 22.1% of NaNs in 68 climbers)
Avg. height in 2022: 164.4 +- 4.9 (74 records, 76 NaN, 50.7% of NaNs in 150 climbers)
Avg. height in 2023: 163.8 +- 5.2 (90 records, 74 NaN, 45.1% of NaNs in 164 climbers)

In [116]:

for year in years:
    df_y = df_f_boulder.loc[df_f_boulder['year'] == year]
    avg_rank = round(df_y['rank'].mean())
    avg_rank_nan = round(df_y.loc[df_y['height'].isnull()]['rank'].mean())
    avg_rank_not_nan = round(df_y.loc[df_y['height'].notnull()]['rank'].mean())
    print(f'{year} avg. rank: {avg_rank}; avg. rank of climbers w/out height {avg_rank_nan},'+
          f' and with height data {avg_rank_not_nan}')

2013 avg. rank: 59; avg. rank of climbers w/out height 59, and with height data 57
2014 avg. rank: 65; avg. rank of climbers w/out height 68, and with height data 46
2015 avg. rank: 43; avg. rank of climbers w/out height 45, and with height data 35
2016 avg. rank: 61; avg. rank of climbers w/out height 65, and with height data 49
2017 avg. rank: 56; avg. rank of climbers w/out height 65, and with height data 37
2018 avg. rank: 52; avg. rank of climbers w/out height 61, and with height data 42
2019 avg. rank: 44; avg. rank of climbers w/out height 59, and with height data 35
2021 avg. rank: 34; avg. rank of climbers w/out height 56, and with height data 28
2022 avg. rank: 75; avg. rank of climbers w/out height 99, and with height data 50
2023 avg. rank: 82; avg. rank of climbers w/out height 107, and with height data 61

In [117]:

outlier_marker = dict(markerfacecolor='none', marker='o', markeredgecolor='orangered', alpha=0.7)
meanprops = dict(markerfacecolor='none', marker='x', markeredgecolor='gray', alpha=0.9, markersize=4)
boxprops=dict(color='darkslateblue')
whiskerprops=boxprops

df_f_boulder.boxplot(column='height', by='year', flierprops=outlier_marker, showmeans=True,
                     meanprops=meanprops, boxprops=boxprops, whiskerprops=whiskerprops)
ax = plt.gca()
ax.grid(linestyle='--', alpha=0.5)
ax.grid(axis='x')
ax.set_yticks(np.arange(151, 182, 3))
plt.suptitle('')
plt.title('Box plot of height distribution of female climbers registered\n'+
          'in the IFSC Bouldering World Cup rankings for different years')
plt.xlabel('year')
plt.ylabel('height [cm]')
plt.show()

Differences between lead and boulder for women¶

In [118]:

mean = df_f_lead_agg['height'].mean()
std = df_f_lead_agg['height'].std()
bins = np.arange(np.floor(df_f_lead_agg['height'].min()), df_f_boulder_agg['height'].max()+2, 1)
plt.hist(df_f_lead_agg['height'], bins=bins, color='darkcyan', zorder=10, alpha=0.55, label='lead')
plt.axvline(mean, color='black', lw=0.95, linestyle='-.')

mean_b = df_f_boulder_agg['height'].mean()
std_b = df_f_boulder_agg['height'].std()
plt.hist(df_f_boulder_agg['height'], bins=bins, color='darkorange', zorder=10, alpha=0.55, label='boulder')
plt.axvline(mean_b, color='black', lw=0.95, linestyle='--')

plt.legend(loc=2)
ax = plt.gca()
ax.set_xticks(bins[::3])
plt.text(0.64, 0.82, f'Avg. height is:\n{round(mean, 1)} $\pm$ {round(std, 1)} in lead\n'+
         f'{round(mean_b, 1)} $\pm$ {round(std_b, 1)} in boulder', transform=ax.transAxes)
plt.title('Height distribution of female climbers registered in\nthe IFSC World Cup rankings between 2013 and 2023')
plt.xlabel('height [cm]')
plt.ylabel('number of climbers')
plt.show()

In [119]:

bf_height_list = np.array(bf_height_list)
bf_height_std_list = np.array(bf_height_std_list)

plt.plot(years, bf_height_list, color='darkorange', lw=1.8, label='boulder')
plt.fill_between(years, bf_height_list - bf_height_std_list, bf_height_list + bf_height_std_list,
                 color='darkorange', lw=0, alpha=0.2)

lf_height_list = np.array(lf_height_list)
lf_height_std_list = np.array(lf_height_std_list)

plt.plot(years, lf_height_list, color='darkcyan', lw=1.8, label='lead')
plt.fill_between(years, lf_height_list - lf_height_std_list, lf_height_list + lf_height_std_list,
                 color='darkcyan', lw=0, alpha=0.2)

plt.xlim([min(years), max(years)])
plt.ylim([150.5, 176])
plt.legend()
ax = plt.gca()
ax.set_xticks(years)
ax.set_yticks(np.arange(151, 176, 2))
plt.title('Comparison of the avg. height of female climbers\nin lead and bouldering IFSC rankings over the years')
plt.xlabel('year')
plt.ylabel('height [cm]')
plt.show()

Drop in height from 2018 to 2019 for lead¶

In [120]:

df_f_lead_18_19 = df_f_lead.loc[(df_f_lead['year'].isin([2018, 2019])) & (df_f_lead['height'].notnull())]
df_f_lead_18_19 = df_f_lead_18_19.groupby(['full_name'], as_index=False)
df_f_lead_18_19 = df_f_lead_18_19.agg({'full_name': 'first', 'country': 'first',
                                       'year': 'sum', 'height': 'mean'})
df_f_lead_18 = df_f_lead_18_19.loc[df_f_lead_18_19['year'] == 2018]
df_f_lead_19 = df_f_lead_18_19.loc[df_f_lead_18_19['year'] == 2019]

print(f'Avg. height of climbers ranked in 2018 but not in 2019: {df_f_lead_18["height"].mean()}')
df_f_lead_18.sort_values('height')

Avg. height of climbers ranked in 2018 but not in 2019: 165.58333333333334

Out[120]:

	full_name	country	year	height
14	dinara fakhritdinova	RUS	2018	152.0
42	manon hily	FRA	2018	154.0
22	gayeon cho	KOR	2018	162.0
25	hsiu-ju lin	TPE	2018	164.0
23	giorgia tesio	ITA	2018	165.0
27	hélène janicot	FRA	2018	165.0
54	patrycja chudziak	POL	2018	168.0
62	tjasa slemensek	SLO	2018	168.0
7	anouck jaubert	FRA	2018	169.0
38	laura stöckler	AUT	2018	172.0
34	katja debevec	SLO	2018	173.0
60	stasa gejo	SRB	2018	175.0

In [121]:

print(f'Avg. height of climbers ranked in 2019 but not in 2018: {df_f_lead_19["height"].mean()}')
df_f_lead_19.sort_values('height')

Avg. height of climbers ranked in 2019 but not in 2018: 161.77777777777777

Out[121]:

	full_name	country	year	height
49	natsuki tanii	JPN	2019	152.0
1	ai mori	JPN	2019	154.0
50	nika potapova	UKR	2019	154.0
65	yuetong zhang	CHN	2019	160.0
39	lucinda ann turnbull	AUS	2019	160.0
4	alma bestvater	GER	2019	161.0
17	elnaz rekabi	IRI	2019	161.0
63	viktoriia meshkova	RUS	2019	161.0
56	risa ota	JPN	2019	161.0
35	kyra condie	USA	2019	162.0
11	chaehyun seo	KOR	2019	163.0
3	alannah yip	CAN	2019	164.0
5	andrea kümin	SUI	2019	164.0
59	shauna coxsey	GBR	2019	164.0
0	afra hönig	GER	2019	165.0
19	eva maria hammelmüller	AUT	2019	167.0
15	elena krasovskaia	RUS	2019	169.0
26	hung ying lee	TPE	2019	170.0

Drop in height from 2018 to 2019 for bouldering¶

In [122]:

df_f_boulder_18_19 = df_f_boulder.loc[(df_f_boulder['year'].isin([2018, 2019])) & (df_f_boulder['height'].notnull())]
df_f_boulder_18_19 = df_f_boulder_18_19.groupby(['full_name'], as_index=False)
df_f_boulder_18_19 = df_f_boulder_18_19.agg({'full_name': 'first', 'country': 'first',
                                             'year': 'sum', 'height': 'mean'})
df_f_boulder_18 = df_f_boulder_18_19.loc[df_f_boulder_18_19['year'] == 2018]
df_f_boulder_19 = df_f_boulder_18_19.loc[df_f_boulder_18_19['year'] == 2019]

print(f'Avg. height of climbers ranked in 2018 but not in 2019: {df_f_boulder_18["height"].mean()}')
df_f_boulder_18.sort_values('height')

Avg. height of climbers ranked in 2018 but not in 2019: 169.1818181818182

Out[122]:

	full_name	country	year	height
47	mina markovic	SLO	2018	161.0
22	giorgia tesio	ITA	2018	165.0
43	megan lynch	USA	2018	167.0
42	maya madere	USA	2018	168.0
56	saki kikuchi	JPN	2018	168.0
61	tjasa slemensek	SLO	2018	168.0
20	franziska sterrer	AUT	2018	169.0
62	vanessa si yinn teng	SGP	2018	169.0
27	isabel gifford	USA	2018	170.0
60	stasa gejo	SRB	2018	175.0
17	emma horan	AUS	2018	181.0

In [123]:

print(f'Avg. height of climbers ranked in 2019 but not in 2018: {df_f_boulder_19["height"].mean()}')
df_f_boulder_19.sort_values('height')

Avg. height of climbers ranked in 2019 but not in 2018: 161.625

Out[123]:

	full_name	country	year	height
36	laura rogora	ITA	2019	152.0
51	natsuki tanii	JPN	2019	152.0
1	ai mori	JPN	2019	154.0
12	camilla moroni	ITA	2019	157.0
48	molly thompson-smith	GBR	2019	159.0
4	alejandra contreras	CHI	2019	160.0
41	mattea pötzi	AUT	2019	160.0
65	yuetong zhang	CHN	2019	160.0
63	viktoriia meshkova	RUS	2019	161.0
55	roxana wienand	GER	2019	162.0
52	naïlé meignan	FRA	2019	164.0
64	vita lukan	SLO	2019	164.0
38	lucia dörffel	GER	2019	165.0
39	lucka rakovec	SLO	2019	170.0
37	laura stöckler	AUT	2019	172.0
59	sienna kopf	USA	2019	174.0

Jump in height from 2013 to 2015 in bouldering¶

In [124]:

df_f_boulder_13_15 = df_f_boulder.loc[(df_f_boulder['year'].isin([2013, 2014, 2015]))
                                      & (df_f_boulder['height'].notnull())]
df_f_boulder_13_15 = df_f_boulder_13_15.groupby(['full_name'], as_index=False)
df_f_boulder_13_15 = df_f_boulder_13_15.agg({'full_name': 'first', 'country': 'first',
                                             'year': lambda x: list(x), 'height': 'mean'})
df_f_boulder_13_15.sort_values('height')

Out[124]:

	full_name	country	year	height
6	dinara fakhritdinova	RUS	[2014, 2013]	152.0
14	jain kim	KOR	[2013]	152.0
19	manon hily	FRA	[2014]	154.0
20	mei kotake	JPN	[2014]	155.0
4	anne-sophie koller	SUI	[2013]	160.0
2	alma bestvater	GER	[2015, 2014, 2013]	161.0
24	risa ota	JPN	[2013]	161.0
7	elnaz rekabi	IRI	[2014]	161.0
22	mina markovic	SLO	[2014, 2013]	161.0
23	petra klingler	SUI	[2015, 2014, 2013]	162.0
18	kyra condie	USA	[2015, 2014, 2013]	162.0
21	miho nonaka	JPN	[2015, 2014]	163.0
13	ievgeniia kazbekova	UKR	[2013]	164.0
25	shauna coxsey	GBR	[2015, 2014, 2013]	164.0
15	julia chanourdie	FRA	[2015, 2013]	164.0
27	tina johnsen hafsaas	NOR	[2015]	164.0
3	andrea kümin	SUI	[2013]	164.0
12	hélène janicot	FRA	[2014, 2013]	165.0
9	fanny gibert	FRA	[2015, 2014, 2013]	165.0
5	chloe caulier	BEL	[2015, 2014, 2013]	166.0
0	akiyo noguchi	JPN	[2015, 2014, 2013]	167.0
10	franziska sterrer	AUT	[2015]	169.0
11	hung ying lee	TPE	[2014]	170.0
1	allison vest	CAN	[2015]	171.0
17	katja debevec	SLO	[2015, 2014]	173.0
16	julija kruder	SLO	[2015, 2014, 2013]	175.0
26	stasa gejo	SRB	[2015]	175.0
8	emma horan	AUS	[2015, 2014]	181.0