import os
import pandas as pd
from sklearn.model_selection import train_test_split

os.system("kaggle datasets download -d tamber/steam-video-games")
os.system("unzip -o steam-video-games.zip")

steam=pd.read_csv('Steam-200k.csv',usecols=[0,1,2,3],names=['userId','game','behavior','hoursPlayed'])
steam.isnull().values.any()
steam['userId'] = steam.userId.astype(str)

print("Zbior danych:")
print(steam)

print("Describe:")
print(steam.describe(include='all'),"\n\n")

print("Gracze z najwieksza aktywnoscia:")
print(steam["userId"].value_counts(),"\n\n")

print("Gracze z najwieksza liczba kupionych gier:")
print(steam[steam["behavior"] != "play"]["userId"].value_counts())
print("Mediana:")
print(steam[steam["behavior"] != "play"]["userId"].value_counts().median(),"\n\n")

print("Gracze ktorzy zagrali w najwieksza liczbe gier:")
print(steam[steam["behavior"] != "purchase"]["userId"].value_counts())
print("Mediana:")
print(steam[steam["behavior"] != "purchase"]["userId"].value_counts().median(),"\n\n")


print("Gry:")
print(steam["game"].value_counts(),"\n\n")

print("Sredni czas grania w grania w dana gre")
print(steam[steam["behavior"] != "purchase"].groupby("game").mean().sort_values(by="hoursPlayed",ascending=False))
print("Mediana:")
print(steam[steam["behavior"] != "purchase"].groupby("game").mean().sort_values(by="hoursPlayed",ascending=False).median(),"\n\n")

print("Najczesciej kupowana gra")
print(steam[steam["behavior"] != "play"]["game"].value_counts())
print("Mediana:")
print(steam[steam["behavior"] != "play"]["game"].value_counts().median(),"\n\n")

print("Gra w ktora zagralo najwiecej graczy")
print(steam[steam["behavior"] != "purchase"]["game"].value_counts())
print("Mediana:")
print(steam[steam["behavior"] != "purchase"]["game"].value_counts().median(),"\n\n")

print("Liczba kupionych gier i liczba gier w ktore gracze zagrali")
print(steam["behavior"].value_counts(),"\n\n")


print("Gra z najwieksza liczba godzin dla jednego gracza")
print(steam[steam["behavior"] != "purchase"][["userId","hoursPlayed","game"]].sort_values(by="hoursPlayed",ascending=False))
print("Mediana:")
print(steam[steam["behavior"] != "purchase"]["hoursPlayed"].sort_values(ascending=False).median(),"\n\n")

print("Suma rozegranych godzin dla danej gry")
print(steam[steam["behavior"] != "purchase"].groupby("game").sum().sort_values(by="hoursPlayed",ascending=False))
print("Mediana:")
print(steam[steam["behavior"] != "purchase"].groupby("game").sum().sort_values(by="hoursPlayed",ascending=False).median(),"\n\n")

#odrzucenie gier dla których jest mniej niż 10 wierszy
steam = steam.groupby("game").filter(lambda x: len(x)>10)
#rozmiar zbioru testowego i dev proporcje 8:1:1
size=int(len(steam)/10)

steam_train, steam_test = train_test_split(steam, test_size=size, random_state=1, stratify=steam["game"])
steam_train, steam_dev = train_test_split(steam_train, test_size=size, random_state=1, stratify=steam_train["game"])

print("Zbior trenujacy")
print(steam_train["game"].value_counts(),"\n")

print("Zbior testujacy")
print(steam_test["game"].value_counts(),"\n")

print("Zbior dev")
print(steam_dev["game"].value_counts(),"\n")