1. testset pred value KDE plot (5FOLD+LGBM)

Y_01.png

Y_02.png

Y_03.png

Y_04.png

Y_05.png

Y_06.png

Y_07.png

Y_08.png

Y_09.png

Y_10.png

Y_11.png

Y_12.png

Y_13.png

Y_14.png

LGBM또한 마찬가지로 비슷한 분포를 가짐을 알 수 있다.

import pandas as pd

import seaborn as sns

import matplotlib.pyplot as plt

fold1 = pd.read_csv('submit_fold1_test.csv')
fold2 = pd.read_csv('submit_fold2_test.csv')
fold3 = pd.read_csv('submit_fold3_test.csv')
fold4 = pd.read_csv('submit_fold4_test.csv')
fold5 = pd.read_csv('submit_fold5_test.csv')

lgbm = pd.read_csv('submit_LightGBM.csv')

# print(lgbm)

for i in range(14):
    sns.kdeplot(fold1[f'Y_{str(i+1).zfill(2)}'], label='fold1')
    sns.kdeplot(fold2[f'Y_{str(i+1).zfill(2)}'], label='fold2')
    sns.kdeplot(fold3[f'Y_{str(i+1).zfill(2)}'], label='fold3')
    sns.kdeplot(fold4[f'Y_{str(i+1).zfill(2)}'], label='fold4')
    sns.kdeplot(fold5[f'Y_{str(i+1).zfill(2)}'], label='fold5')
    
    sns.kdeplot(lgbm[f'Y_{str(i+1).zfill(2)}'], label='LGBM')
    plt.legend()
    plt.savefig(f'Y_{str(i+1).zfill(2)}.png')
    plt.close()

2. what’s diffrence with Ground truth, Prediction? (cross validation)

fold1 Y_01.png

fold1 Y_02.png

fold1 Y_03.png

fold1 Y_04.png

fold1 Y_05.png

fold1 Y_06.png