import pandas as pd
from sklearn.model_selection import train_test_split

# get data
sells = pd.read_csv('data/Property Sales of Melbourne City.csv')

# delete unnecessary columns and drop rows with NaN values
columns_to_drop = [
  'Lattitude',
  'Longtitude',
  'CouncilArea',
  'Propertycount',
  'Method',
  'SellerG',
  'Date',
  'Postcode',
  'Bedroom2',
  'Bathroom',
  'Car',
  'BuildingArea',
  'Address'
  ]
sells = sells.drop(columns_to_drop, axis=1).dropna()

# normalize values
sells["Price"] = sells["Price"] / sells["Price"].max()
sells["Landsize"] = sells["Landsize"] / sells["Landsize"].max()
sells["Distance"] = sells["Distance"] / sells["Distance"].max()

# split to train/dev/test subsets
X = sells
Y = sells.pop('Price')

X_train, X_temp, Y_train, Y_temp = train_test_split(X, Y, test_size=0.3, random_state=1)
X_val, X_test, Y_val, Y_test = train_test_split(X_temp, Y_temp, test_size=0.5, random_state=1)

# save subsets to files
X_train.to_csv('X_train.csv', index=False)
X_val.to_csv('X_val.csv', index=False)
X_test.to_csv('X_test.csv', index=False)

Y_train.to_csv('Y_train.csv', index=False)
Y_val.to_csv('Y_val.csv', index=False)
Y_test.to_csv('Y_test.csv', index=False)