Lianne & Justin @ Just into Data liannewriting

## print-best-estimator.py
opt.best_estimator_.steps

## print-best-estimator.py
opt.best_estimator_

## fit_xgboost.py
opt.fit(X_train, y_train)

## explore_data.py
df.info()

df['result'].value_counts()

## pmdarima_auto_arima.py
import pmdarima as pm
auto_arima = pm.auto_arima(df_train, stepwise=False, seasonal=False)
auto_arima

## train_test_split.py
msk = (df.index < len(df)-30)
df_train = df[msk].copy()
df_test = df[~msk].copy()

## evaluation_manual.py
from sklearn.metrics import mean_absolute_error, mean_absolute_percentage_error, mean_squared_error

mae = mean_absolute_error(df_test, forecast_test)
mape = mean_absolute_percentage_error(df_test, forecast_test)
rmse = np.sqrt(mean_squared_error(df_test, forecast_test))

print(f'mae - manual: {mae}')
print(f'mape - manual: {mape}')
print(f'rmse - manual: {rmse}')

## prediction_manual_auto_comparison.py
forecast_test_auto = auto_arima.predict(n_periods=len(df_test))
df['forecast_auto'] = [None]*len(df_train) + list(forecast_test_auto)

df.plot()

## evaluation_auto.py
mae = mean_absolute_error(df_test, forecast_test_auto)
mape = mean_absolute_percentage_error(df_test, forecast_test_auto)
rmse = np.sqrt(mean_squared_error(df_test, forecast_test_auto))

print(f'mae - auto: {mae}')
print(f'mape - auto: {mape}')
print(f'rmse - auto: {rmse}')

## auto_arima_summary.py
auto_arima.summary()
	import pmdarima as pm
	auto_arima = pm.auto_arima(df_train, stepwise=False, seasonal=False)
	auto_arima
	msk = (df.index < len(df)-30)
	df_train = df[msk].copy()
	df_test = df[~msk].copy()
	from sklearn.metrics import mean_absolute_error, mean_absolute_percentage_error, mean_squared_error

	mae = mean_absolute_error(df_test, forecast_test)
	mape = mean_absolute_percentage_error(df_test, forecast_test)
	rmse = np.sqrt(mean_squared_error(df_test, forecast_test))

	print(f'mae - manual: {mae}')
	print(f'mape - manual: {mape}')
	print(f'rmse - manual: {rmse}')
	forecast_test_auto = auto_arima.predict(n_periods=len(df_test))
	df['forecast_auto'] = [None]*len(df_train) + list(forecast_test_auto)

	df.plot()