We are working on time series data forecasting. Our input data set is large, so thought to average two consecutive data points and reduce it to half the size. But we have observed that average values increased slope of the line, which effecting our forecast values slop too.

Here is screenshot of the issue I am describing. (Please NOTE that this is dummy data I quickly put in to describe the problem).

I understand that increase in slope value for summarized data make sense because AVG values are higher comparing with originals, the issue I am struggling with (or) trying to answer is, is there any way we can summarize the data without increasing the slop to keep forecast within original range?

enter image description here

Thank you for your time!


1 Answer 1


Plain and simple: Include the time information when plotting your data and calculating your slope.

Right now, you use the following data for this:

1   150446.5
2   150488.5
3   150530.5
4   150613.5
5   150613.5

What you should use is however:

1.5 150446.5
3.5 150488.5
5.5 150530.5
7.5 150613.5
9.5 150613.5

Note how the first time value in each column corresponds to the average of the times corresponding to the values which you used to calculate the respective mean.

  • $\begingroup$ Thanks for the response! It is not just simple display purpose, I am doing some programming. Are there any proven algorithms for this? $\endgroup$
    – kosa
    Commented Nov 13, 2014 at 14:59
  • $\begingroup$ @Nambari: This applies to programming as well, in particular calculating your slope. You always have to use the appropriate times. (And your wrong results perfectly match what you get when you do not do so.) $\endgroup$
    – Wrzlprmft
    Commented Nov 13, 2014 at 15:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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