0
votes

I'm using R(3.1.1), and NNetar models for forecasting. I would like to know what should be the "frequency" parameter, which is assigned in the ts() function, if I'm using time series data which is:

Separated by 15 minutes and is spread over 365 days starting from 01 April 2018 to 31 Mar 2019.

Dataset example: Date Time Value 01/04/2018 00:00 49.99 01/04/2018 00:15 50.01 01/04/2018 00:30 50.03.....

2

2 Answers

0
votes

Aggregate your 15 min data into 60 min (hourly) or daily basis then try to forecast.

0
votes

Since you are having 15min interval data, you can check for seasonality what you want to consider

If your data consist of daily seasonality then frequency = 96 or weekly then frequency = 672.

If you think you need to incorporate both the seasonality then you can transform with msts() in R.