Denysyefimov

denysyefimov_code/task1_titanic_houseprice/denysyefimov_houseprices.py

@@ -0,0 +1,142 @@
+from sklearn.metrics import mean_squared_error
+from sklearn import linear_model
+import pandas as pd
+import matplotlib.pyplot as plt
+from sklearn.model_selection import train_test_split
+
+
+def one_hot_encode(x):
+    return 1 if x == 'Partial' else 0
+
+
+def encode1(x):
+    return 3 if x == 4 else x
+
+
+if __name__ == '__main__':
+
+    # add data in file
+
+    train = pd.read_csv('../input/house-prices-advanced-regression-techniques/train.csv')
+    test = pd.read_csv('../input/house-prices-advanced-regression-techniques/test.csv')
+
+    # making some changes
+
+    deviation = np.log(train.SalePrice)
+
+    # Now let's look at what influences
+    # the price of the house itself and analyze the data.
+
+    influences = train.select_dtypes(include=[np.number])
+    influences.dtypes
+    percent = influences.corr()
+    print(percent['SalePrice'].sort_values(ascending=False)[:5], '\n')
+    print(percent['SalePrice'].sort_values(ascending=False)[-5:])
+    quality_pivot = train.pivot_table(index='OverallQual',
+                                      values='SalePrice',
+                                      aggfunc=np.median
+                                      )
+    print(quality_pivot)
+
+    # Thus, we got that the price is most
+    # affected by: OverallQual, GrLivArea, GarageCars,
+    # GarageArea, YrSold, OverallCond,
+    # MSSubClass, EnclosedPorch, KitchenAbvGr.
+    # Now let's see the rest of the
+    # factors that affect the price.
+    # Thus, we got the graphs and now
+    # it would be worth removing the points that could
+    # shift our line from the "center
+    # of events".
+    # Let's set the maximum value of
+    # the garage area to 1200, so that the line
+    # doesn't run away from us.
+    # And makes some changes in GrLivArea
+
+    train = train[train['GarageArea'] < 1200]
+    train = train[train['GrLivArea'] < 4000]
+
+    # Now let's check all NULL
+
+    nulls = pd.DataFrame(train.isnull().sum().sort_values(ascending=False)[:25])
+    print(nulls)
+
+    # Now let's chek data without numbers(non-numeric)
+
+    categoricals = train.select_dtypes(exclude=[np.number])
+    categoricals.describe()
+
+    # Let's use one-hot encoding method for our non-numeric data
+
+    train['enc_street'] = pd.get_dummies(train.Street,
+                                         drop_first=True
+                                         )
+    test['enc_street'] = pd.get_dummies(train.Street,
+                                        drop_first=True
+                                        )
+
+    train['enc_condition'] = train.SaleCondition.apply(one_hot_encode)
+    test['enc_condition'] = test.SaleCondition.apply(one_hot_encode)
+    train['enc_condition1'] = train.KitchenAbvGr.apply(encode1)
+    test['enc_condition1'] = test.KitchenAbvGr.apply(encode1)
+
+    data = train.select_dtypes(include=[np.number]).interpolate().dropna()
+
+    # Thus, we redid the reasons for the sale and
+    # got a simpler model, which is easier to work
+    # with both for us
+    # and the computer. And also, those data that
+    # we do not use have been converted to zero,
+    # also in order to
+    # make it easier for us to work.
+    # Now let's start building our linear model.
+
+    y = np.log(train.SalePrice)
+    X = data.drop(['SalePrice', 'Id'], axis=1)
+    X_train, X_test, y_train, y_test = train_test_split(X,
+                                                        y,
+                                                        random_state=100,
+                                                        test_size=.33
+                                                        )
+    lr = linear_model.LinearRegression()
+    model = lr.fit(X_train, y_train)
+    predictions = model.predict(X_test)
+    actual_values = y_test
+
+    # Now let's build our line with a predicted price.
+    # Now let's change our alpha a little and get some
+    # graphics. With alpha = 1 we get better prediction.
+
+    alpha = 1
+    rm = linear_model.Ridge(alpha=alpha)
+    ridge_model = rm.fit(X_train, y_train)
+    preds_ridge = ridge_model.predict(X_test)
+    plt.scatter(preds_ridge,
+                actual_values,
+                alpha=.75,
+                color='black'
+                )
+    plt.xlabel('Predicted Price')
+    plt.ylabel('Actual Price')
+    plt.title('Ridge Regularization with alpha = {}'.format(alpha))
+    overlay = 'R^2 is: {}\nRMSE is: {}'.format(ridge_model.score(X_test, y_test),
+                                               mean_squared_error(y_test, preds_ridge))
+    plt.annotate(s=overlay,
+                 xy=(12.1, 10.6),
+                 size='x-large'
+                 )
+
+    # Finally, we get our final answer.
+
+    submission = pd.DataFrame()
+    submission['Id'] = test.Id
+    feats = test.select_dtypes(include=[np.number]).drop(['Id'],
+                                                         axis=1
+                                                         ).interpolate()
+    predictions = ridge_model.predict(feats)
+    final_predictions = np.exp(predictions)
+    print("Final predictions are: \n", final_predictions[:5])
+    submission['SalePrice'] = final_predictions
+    submission.head()
+    submission.to_csv('submission1.csv', index=False)
+    print("R^2 is: \n", model.score(X_test, y_test))