Time Series Forecasting for irregular time series in R



Hi…I want to implement Time Series Forecasting for irregular time series(i.e it have irregular interval in dates) on Daily Dataset. My task is to find out the best time series forecasting models which can make forecasting for the number of request for the next 15 days.
The example of the dataset which I am using is given below:
date req_per_day
13-07-2016 1
15-07-2016 1
11-08-2016 1
01-09-2016 1
13-09-2016 1
14-09-2016 1
22-09-2016 2

For that i have used “zoo” library in R for the irregular time series but I am unable to implement ets and other methods like ar model,arima model with zoo objects as it expects the time series to be regular.I have also tried in converting the zoo object to ts but again it is showing NA for the dates which are not present in the dataset.

My question is what is the method we use for forecasting daily data with gaps in it in R.

Thanks in advance.


hi @sonam09

do you want to stock to Arima ? if not try the great new package from Facebook prophet, It handles NA and exceptions. You can even give period of exceptions

Best regards


Thanks…I will use this library for the time series forecasting.


I am not a savy in teimseries, however I believe if you try using any imputation methods to fill the missing values (hopefully you will have more data in your dataset) it will suffice your requirements.

normally in R , implementing the time series is convenient, i believe you are converting the independent variable(Date) to factor before passing it to any time series models.

I will try getting a exact code from R sooner for this problem.