53 lines
1.8 KiB
Python
53 lines
1.8 KiB
Python
|
import pandas as pd
|
||
|
from sklearn.model_selection import train_test_split
|
||
|
from sklearn import preprocessing
|
||
|
import kaggle
|
||
|
|
||
|
kaggle.api.authenticate()
|
||
|
|
||
|
kaggle.api.dataset_download_files('ruchi798/movies-on-netflix-prime-video-hulu-and-disney', path='.', unzip=True)
|
||
|
|
||
|
# odczyt danych
|
||
|
film_data = pd.read_csv('MoviesOnStreamingPlatforms_updated.csv')
|
||
|
|
||
|
# Czyszczenie wierszy z pustymi warościami.
|
||
|
film_data.dropna(inplace=True)
|
||
|
|
||
|
# Usunięcie zbędnych kolumn
|
||
|
film_data.drop(film_data.columns[[0, 1]], axis = 1)
|
||
|
|
||
|
# Normalizacja: Lowercase dla danych tekstowych, standaryzacja (0..1) dla wartości float, sortowanie danych w komórce.
|
||
|
|
||
|
for col_name in ['Title', 'Directors', 'Genres', 'Country', 'Language']:
|
||
|
film_data[col_name] = film_data[col_name].str.lower()
|
||
|
|
||
|
for col_name in ['Directors', 'Genres', 'Country', 'Language']:
|
||
|
film_data[col_name] = film_data[col_name].str.split(',').map(lambda x: ','.join(sorted(x)))
|
||
|
|
||
|
scaler = preprocessing.MinMaxScaler()
|
||
|
film_data[['IMDb', 'Runtime']] = scaler.fit_transform(film_data[['IMDb', 'Runtime']])
|
||
|
|
||
|
# Podział zbioru na train, dev, test w proporcji 8:1:1
|
||
|
train_ratio = 0.8
|
||
|
validation_ratio = 0.1
|
||
|
test_ratio = 0.1
|
||
|
|
||
|
film_train, film_test = train_test_split(film_data, test_size=1 - train_ratio)
|
||
|
|
||
|
film_valid, film_test = train_test_split(film_test, test_size=test_ratio/(test_ratio + validation_ratio))
|
||
|
|
||
|
# Statystki głównego zbioru i podzbiorów
|
||
|
for i, data_set in enumerate([film_data, film_train, film_valid, film_test]):
|
||
|
if i == 0:
|
||
|
print("Główny zbiór danych")
|
||
|
elif i == 1:
|
||
|
print("Zbiór trenujący")
|
||
|
elif i == 2:
|
||
|
print("Zbiór walidujący")
|
||
|
if i == 3:
|
||
|
print("Zbiór testowy")
|
||
|
print(len(data_set))
|
||
|
print(data_set.describe().loc[['count','mean', 'max', 'min', 'std', '50%']])
|
||
|
[print(data_set[name].value_counts()) for idx, name in enumerate(data_set)]
|
||
|
|