Hello,
I have a set of 21 time-series with weekly data of the same variable from years 2000 to 2020 (January to mid-May only). The figure illustrates those 21 time-series.
I want to evaluate if one of these time-series (year 2020, black line) is statistically different than the set of other 20 time-series (years 2000 to 2019, colored lines) or if its variation is random noise within the dataset.
What would be the proper statistical test for this comparison?
Thanks.
Hi well you can compare the autocorrelation functions. You can also use Dynamic time warping. Other possible methods are using a measure on entropy based in metrics or Motif Discovery.
I hope that helps.
Paul Louangrath – your link allows someone to download an illegal copy of a copyright textbook. You should edit your post and remove it.
Gopal Kanji worked hard to have his book published, and he deserves his (probably meagre!) royalties.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics