Sonam Tripathi
2017-Apr-19 09:03 UTC
[R] Time Series forecasting models for irregular time series for zoo objects
Hi I am new with R.Currently I am using time series forecasting to do daily
forecasting for predicting request per day.The dataset which i am using has
unevenly spaced date a snapshot of dataset is given below I have to find
the best model which can help me in predicting the future request which can
make forecast for next 15 days.I have used zoo package for dealing with
irregular time series data but now I am unable to implement "ets" and
"ar"
and "arima" model with this as it expects the series to be regular and
evenly spaced and it gives me warning " Missing values encountered. Using
longest contiguous portion of time series" but still I proceed for
forecast then it is showing me date in the format like "1.711600e+04"
and
point of forecast is same for nearly all the dates.
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
23-09-2016 1
26-09-2016 2
27-09-2016 4
29-09-2016 2
My question is how to implement ets and other model on zoo objects which
show me correct date format.Based on other suggestions i have also tried
converting zoo to ts but it is again showing "NA" for the dates which
are
not present in the dataset and give same result which I specified above.
How to do time series forecasting with uneven or irregular dates using zoo
objects in R.