2022-04-02 20:28:16 +02:00
|
|
|
import pandas as pd
|
|
|
|
|
|
|
|
print('--Full dataset stats--')
|
2022-04-02 20:47:36 +02:00
|
|
|
cars = pd.read_csv('./Car_Prices_Poland_Kaggle.csv', encoding="utf-8")
|
2022-04-02 20:28:16 +02:00
|
|
|
print(cars.describe(include='all'))
|
|
|
|
|
|
|
|
print('Dev dataset stats')
|
2022-04-02 20:47:36 +02:00
|
|
|
cars_dev = pd.read_csv('./Car_Prices_Poland_Kaggle_dev.csv', encoding="utf-8")
|
2022-04-02 20:28:16 +02:00
|
|
|
print(cars_dev.describe(include='all'))
|
|
|
|
|
|
|
|
print('# statystyki dla zbioru test')
|
2022-04-02 20:47:36 +02:00
|
|
|
cars_test = pd.read_csv('./Car_Prices_Poland_Kaggle_test.csv', encoding="utf-8")
|
2022-04-02 20:28:16 +02:00
|
|
|
print(cars_test.describe(include='all'))
|
|
|
|
|
|
|
|
print('# statystyki dla zbioru train')
|
2022-04-02 20:47:36 +02:00
|
|
|
cars_train = pd.read_csv('./Car_Prices_Poland_Kaggle_train.csv', encoding="utf-8")
|
2022-04-02 20:28:16 +02:00
|
|
|
print(cars_train.describe(include='all'))
|