import pandas as pd
from sklearn.ensemble import RandomForestRegressor

df = pd.read_csv('../DATA/Advertising.csv')

df

df.describe()

X = df.drop('sales',axis=1)
y = df['sales']

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.30, random_state=101)

# HOLD OUT SET

# Further split 30% of test into validation and hold-out (15% and 15% each)
X_validation, X_holdout_test, y_validation, y_holdout_test = train_test_split(X_test, y_test, test_size=0.5, random_state=101)

model = RandomForestRegressor(n_estimators=10,random_state=101)

model.fit(X_train,y_train)

validation_predictions = model.predict(X_validation)

from sklearn.metrics import mean_absolute_error,mean_squared_error

mean_absolute_error(y_validation,validation_predictions)

mean_squared_error(y_validation,validation_predictions)**0.5 #RMSE

model = RandomForestRegressor(n_estimators=35,random_state=101)
model.fit(X_train,y_train)

validation_predictions = model.predict(X_validation)

mean_absolute_error(y_validation,validation_predictions)

mean_squared_error(y_validation,validation_predictions)**0.5 #RMSE

model = RandomForestRegressor(n_estimators=35,random_state=101)
model.fit(X_train,y_train)

test_predictions = model.predict(X_holdout_test)

mean_absolute_error(y_holdout_test,test_predictions)

mean_squared_error(y_holdout_test,test_predictions)**0.5

final_model = RandomForestRegressor(n_estimators=35,random_state=101)

final_model.fit(X,y)

import joblib

joblib.dump(final_model,'final_model.pkl')

X.columns

list(X.columns)

joblib.dump(list(X.columns),'column_names.pkl')

col_names = joblib.load('column_names.pkl')

col_names

loaded_model = joblib.load('final_model.pkl')

loaded_model.predict([[230.1,37.8,69.2]])