%run Utility_tools.ipynb

%run Libraries.ipynb

raw = pd.read_csv('diamond_pred.csv', index_col=[0]) df = deepcopy(raw)

df.sample(4)

df.drop(['x', 'y', 'z', 'table', 'clarity', 'depth', 'color', 'cut'], axis=1, inplace=True)

df = pd.concat([df.iloc[:,-1], df.iloc[:,:-1]], axis=1)

#Pre-modeling x = df.iloc[:,:-1] y = df.iloc[:,-1] x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.20, random_state=42) scaler = RobustScaler() x_train = scaler.fit_transform(x_train) x_train = pd.DataFrame(x_train) scaler = RobustScaler() x_test = scaler.fit_transform(x_test) x_test = pd.DataFrame(x_test)

lin_mod = LinearRegression() lin_mod.fit(x_train, y_train) preds = lin_mod.predict(x_test)

rmse = np.sqrt(mean_squared_error(y_test, preds)) rmse

lin_mod.score(x_train, y_train)

r2_score(y_test, preds)

mean_absolute_error(y_test, preds)

normalized_rmse = rmse/(df['price'].max()-df['price'].min()) #rmse divided by the target's range normalized_rmse

df_1 = deepcopy(raw)

df_1.drop(['size_volume', 'table', 'clarity', 'depth', 'color', 'cut'], axis=1, inplace=True)

df_1 = pd.concat([df_1.iloc[:,2:], df_1.iloc[:,:2]], axis=1)

#Pre-modeling x = df_1.iloc[:,:-1] y = df_1.iloc[:,-1] x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.20, random_state=42) scaler = RobustScaler() x_train = scaler.fit_transform(x_train) x_train = pd.DataFrame(x_train) scaler = RobustScaler() x_test = scaler.fit_transform(x_test) x_test = pd.DataFrame(x_test)

lin_mod = LinearRegression() lin_mod.fit(x_train, y_train) preds = lin_mod.predict(x_test)

rmse = np.sqrt(mean_squared_error(y_test, preds)) rmse

lin_mod.score(x_train, y_train)

r2_score(y_test, preds)

mean_absolute_error(y_test, preds)

normalized_rmse = rmse/(df['price'].max()-df['price'].min()) normalized_rmse

df_2 = deepcopy(raw)

df_2 = df_2[['carat', 'x', 'price']]

#Pre-modeling x = df_2.iloc[:,:-1] y = df_2.iloc[:,-1] x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.20, random_state=42) scaler = RobustScaler() x_train = scaler.fit_transform(x_train) x_train = pd.DataFrame(x_train) scaler = RobustScaler() x_test = scaler.fit_transform(x_test) x_test = pd.DataFrame(x_test)

lin_mod = LinearRegression() lin_mod.fit(x_train, y_train) preds = lin_mod.predict(x_test)

rmse = np.sqrt(mean_squared_error(y_test, preds)) rmse

lin_mod.score(x_train, y_train)

r2_score(y_test, preds)

mean_absolute_error(y_test, preds)

normalized_rmse = rmse/(df['price'].max()-df['price'].min()) normalized_rmse

df_3 = deepcopy(raw)

df_3.sample(3)

df_3_cat = df_3.select_dtypes(include='object')

df_3 = pd.concat([pd.get_dummies(df_3_cat), df_3.select_dtypes(include='number')], axis=1)\ .drop(['depth', 'table', 'size_volume'], axis=1)

df_3 = pd.concat([df_3[[column for column in df_3.columns if column!='price']], df_3['price']], axis=1)

#Pre-modeling x = df_3.iloc[:,:-1] y = df_3.iloc[:,-1] x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.20, random_state=42) scaler = RobustScaler() x_train = scaler.fit_transform(x_train) x_train = pd.DataFrame(x_train) scaler = RobustScaler() x_test = scaler.fit_transform(x_test) x_test = pd.DataFrame(x_test)

lin_mod = LinearRegression() lin_mod.fit(x_train, y_train) preds = lin_mod.predict(x_test)

rmse = np.sqrt(mean_squared_error(y_test, preds)) rmse

lin_mod.score(x_train, y_train)

r2_score(y_test, preds)

mean_absolute_error(y_test, preds)

normalized_rmse = rmse/(df_3['price'].max()-df_3['price'].min()) normalized_rmse #We get the best result when using categorical features as well

rfc_ = RandomForestRegressor(random_state=42)

param_grid = { 'n_estimators': [300, 500], 'max_features': ['auto', 'sqrt', 'log2'], 'max_depth' : [12], "min_samples_split" : [2,4,8], "bootstrap": [True, False] }

CV_rfc = GridSearchCV(estimator=rfc_, param_grid=param_grid, cv=5) CV_rfc.fit(x_train, y_train)

CV_rfc.best_params_

rfc1=RandomForestClassifier(n_estimators=, max_features= , max_depth=, min_samples_split=, bootstrap=) rfc1.fit(x_train, y_train)

rf_pred = rfc1.predict(x_test)

plot_rf_feat_importance(rfc1, x)

rmse = np.sqrt(mean_squared_error(y_test, rf_pred) print(f'''Root Mean Squared Error: {rmse}''') print(f'''R^squared on train: {rfc1.score(x_train, y_train)}''') print(f'''R^squared on test: {r2_score(y_test, rf_pred)}''') print(f'''Mean Absolute Error: {mean_absolute_error(y_test, rf_pred)}''') print(f'''Normalized Root Mean Squared Error: {rmse/(df_3['price'].max()-df_3['price'].min())}