0

I am dealing with a dataset that has different input data with a different behavior but I am getting confused as I can not find any trend or pattern in my data. I checked seasonality and found nothing. I checked the literature for data that does not have trend and seasonality, but I do not know exactly what is the proper treatment for prediction and interpretation of the results. What is the solution when the data is like this?

I also plotted a heatmap correlation matrix and to me, this matrix says that I am dealing with data that are not correlated. Am I right about these two concepts? If so what should be done as further steps?

If you think the figure is not good to be pasted in the question please do not downvote and just let me know, I will promise to solve it. Thanks.

enter image description here

Richard Hardy
  • 67,272
john22
  • 127
  • 2
    What decomposition tool are you using that thinks that the signal in your time series is entirely trend, and has residuals that are apparently all 1?? To your question, many time series have neither trend nor seasonality. What this means can only be determined through domain knowledge. Find someone who can tell you where the three large spikes came from, as well as the other changes in your time series. Take a look at this: https://stats.stackexchange.com/q/222179/1352 – Stephan Kolassa Mar 07 '23 at 14:54
  • 1
    The first step is to find software that works correctly! In your case, evidently something changed around time 700. You will need to think about the meaning of this series and consider whether it is helpful to use data before that change for predicting subsequent behavior. – whuber Mar 07 '23 at 15:24

0 Answers0