Rookie Seasons

import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

df = pd.read_csv('Rookie Seasons.csv',index_col='Rk')
u21 = pd.read_csv('U21.csv')
arm = pd.read_csv('arm_strength.csv')
tools = pd.read_csv('5tools.csv')
tools = tools.merge(arm,how='left',on='player_id')
tools = tools.dropna(subset=['n_outs_above_average','sprint_speed','arm_overall'])
#print(tools.corr())
print(tools.quantile(q=.66))
tools = tools[tools['batting_avg']>=.28]
#tools = tools[tools['on_base_plus_slg']>=.83176]
tools = tools[tools['slg_percent']>.486]
tools = tools[tools['n_outs_above_average']>=1]
tools = tools[tools['sprint_speed']>=28.2]
tools = tools[tools['arm_overall']>=87.776]
tools = tools[tools['exit_velocity_avg']>=89.9]
tools

julio = u21[u21['Player']=='Julio Rodríguez']
julio

import matplotlib.style as style
style.use('fivethirtyeight')
sns.catplot(data=u21,x='BA',kind='box',color='aliceblue',saturation=1,aspect=2,fliersize=0)
sns.stripplot(data=u21,x='BA',size=7,jitter=0.2)
sns.stripplot(data=julio,x='BA',size=14,color='mediumseagreen')
plt.xlabel('Batting Average')
plt.show()



u21