1
$\begingroup$

I have a dataset for weekly number of calls to a call center for three years.The data is seasonal (I know this from practitioners knowledge) which means that calls normally come on summer and winter. However, without having practitioners knowledge, how can I check if my data is intermittent or this is seasonality which is impacting zeros for specific time intervals. Is there anyway in R that I can do that? The reason is that I need to do further analysis on those parts of the data which are not zero and I need to classify my data to in-season and off-season. Any help is appreciated.

$\endgroup$
3
  • $\begingroup$ why don't you post your data . Daily data is best as what happens in week i of year 1 may not happen in week i of year 2 etc. while the analysis of daily data can pick up the kind of structure that you are after. $\endgroup$ – IrishStat Apr 21 '15 at 15:18
  • $\begingroup$ I dont have daily data. It is weekly and it barely take values. So daily data is going to bring much more zeros to my data set. I also dont need accurate forecasts for daily data I just need to know my dataset and the inherited trend in that. $\endgroup$ – Fairy Apr 21 '15 at 15:29
  • $\begingroup$ OK .... please post your data and some of our experts in time series analysis will give you some answers. $\endgroup$ – IrishStat Apr 21 '15 at 15:40
1
$\begingroup$

You can decompose (STL decomposition) you time series into trend, seasonal and remainder using "forecast" package. See example and explanation in the amaizing online book by Rob J Hyndman and George Athana­sopou­los, who are also the authors of the package. Once decomposed, you can analyze trend or remainder or both.

$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.