Exercise

# Build time series forecast for new product

Before we can even calculate a bottom-up forecast for the metropolitan region we need to have forecasts of multiple products! First, let's build a time series forecast of the specialty product in the metropolitan region. The product demand is saved as `MET_sp`

in your workspace as well as `dates_valid`

as well as your validation data `MET_sp_v`

.

You've written the MAPE function enough at this point. A `mape()`

function has now been written for you to use with two inputs: the first is the forecast and the second is the validation set.

Instructions

**100 XP**

- Use the
`auto.arima()`

function to build a time series model for the specialty product`MET_sp`

. - Forecast this model for 22 time periods into 2017.
- Make this forecast into an
`xts`

object. You can still use the`dates_valid`

object for the`order.by =`

option. - Calculate the MAPE for this forecast with your new
`mape()`

function.