R Data Analysis Projects

Deep networks for time series prediction

If you remember the initial regression problem for time series forecasting, we used the previous kings' ages at the time of their deaths to predict the next king's age at his time of death. We are going to do something similar here.

We are going to use a rolling window approach.

Our first training record will be constructed as follows:

• The closing price from day 1 in our dataset, up to day 31, will be our X. The day 32 closing price will be our Y.

Now, our second training record will be as follows:

• The closing price from day 2 in our dataset, up to day 32, will be our X. The closing price on day 33 will be our Y.

We will follow the same pattern. Once again, we will leverage quantmod package's ...