0

How would you do regression to predict the sate in a future time:

       SeriesYear  MonthDay             State
0        1  2019    12  13   [1, 0, 0, 1, 0, 0]
1        2  2019    12  17   [0, 1, 0, 0, 1, 0]
2        3  2019    12  20   [0, 0, 1, 0, 1, 0]
3        4  2019    12  24   [0, 1, 0, 1, 0, 0]
4        5  2019    12  27   [0, 1, 0, 0, 1, 0]
5        6  2019    12  31   [0, 0, 0, 1, 0, 1]
6        7  2020     1   3   [1, 0, 0, 0, 0, 1]
.
.
.       some future date                 ?

Basically I want to know state in some future time in the form of a binary list?

NOTE:

Every single row has its own unique state that is not the same as any other row.

Rebel
  • 472
  • 8
  • 25
  • It doesn't sound like you want to do regression, but more like classification. I'd check to see how many unique `State` values I have and if it's feasible then I'd just treat each unique `State` as a classification label. I suppose you could also treat each element in `State` as its own label and perform multi-class classification. – Sean Feb 05 '23 at 00:13
  • Thank you Sean, I just looked up my table. Every single state in the rows is unique. That is unfortunate that I cannot establish a predictive model. Right? – Rebel Feb 05 '23 at 00:33
  • Are you saying that you have exactly one sample for each `State`? If that's the case, then no using the `State` elements themselves as labels is not viable. – Sean Feb 05 '23 at 01:07
  • yes, that's exactly the case. – Rebel Feb 05 '23 at 01:08

0 Answers0