In this question, we will try to use time series model such as ARIMA and others to build the model(s) for forecasting the future. Create the acf and pacf plots for df_day dataframe. Find the best model with different parameters on ARIMA model. The parameter range for p,d,q are all from [0, 1, 2]. In total, you need to find out the best model with lowest Mean Abosulate Error from 27 choices based on the time from "Jul-01-2014" to "Dec-01-2014". Using the best model in above steps to forecast the time from "Jan-01-2015" to "Jan-31-2015". Plot the predicted value and the true demand value from "Jan-01-2015" to "Jan-31-2015". Could you think of any other model (not as same as ARIMA) could do the forecasting for demand value from "Jan-01-2015" to "Jan-31-2015"? You could choose one model (except ARIMA) and train the model based on the demand value from "Jul-01-2014" to "Dec-01-2014" (same training data as the ARIMA). Hint: there are some resources regarding other time series forecasting models such as prophet here and also the exponential smoothing here