If my dataset is not a time series with uniform intervals, will the models still be effective?

[eotich32]

For instance, if my data looks like :

ds y
2024-10-01 y1
2024-10-02 y2
2024-10-05 y3
2024-10-14 y4
2024-10-16 y5
2024-10-19 y6
....

[marcopeix]

Hello! The main problem will be in setting the correct frequency to create the dates, as we only support valid pandas or polars offset alias. In this case, I suggest you use integers instead of dates and specify freq=1. The model will still work, since internally, it tries to map an input sequence to an output sequence.

Let me know if that answers your question!

[eotich32]

Hello! The main problem will be in setting the correct frequency to create the dates, as we only support valid pandas or polars offset alias. In this case, I suggest you use integers instead of dates and specify freq=1. The model will still work, since internally, it tries to map an input sequence to an output sequence.

Let me know if that answers your question!

It's great, this method solve my problem, thank you!

[marcopeix]

You're welcome!