import pandas as pd
import numpy as np

df0 = pd.read_csv('data/stat0.csv')

def process(df, val_col):
    dfr = df.loc[:,['name', val_col]].sort_values(by=['name', val_col]).groupby('name', as_index=False).apply(lambda grp: grp.iloc[2:-2])
    dfr.to_csv('data/{}-percentiled.csv'.format(val_col), index=False)
    return dfr
    
process(df0, 'all').boxplot(by='name')

def process_agg(df, val_col):
    dfr = df.loc[:,['name', val_col]].groupby('name', as_index=False).apply(lambda grp: grp.iloc[2:-2])
    dfr = dfr.groupby('name').agg([np.min, np.median, np.max])
    dfr.columns = dfr.columns.droplevel(0)
    return dfr.rename(columns={'amin': 'min', 'amax': 'max'})

process_agg(df0, 'all').plot()