Survey banner
The Dataiku Community is moving to a new home! We are temporary in read only mode: LEARN MORE

Time series extrapolation with at least one row

Solved!
EdBerth
Level 3
Time series extrapolation with at least one row

I need to calculate the rolling average of orders by customers.

I'm working with the Time Series resampling recipe.

I have sometimes only one order for customers and in this case the extrapolation, with constant value = 0 after the order date, is not calculated.

 

how to have a calculated extrapolation with at least one row ?

0 Kudos
1 Solution
StanG
Dataiker

Hi,

Series with only 1 row are ignored because interpolation on them is not possible, this is a known limitation. In our case, I'll advise you to manually create 2 extra rows for the first and last date with 0 as value for each series and select "no extrapolation" (but keep interpolation) to be sure they are all interpolated between these 2 dates only. 

View solution in original post

0 Kudos
3 Replies
StanG
Dataiker

Hi,

Series with only 1 row are ignored because interpolation on them is not possible, this is a known limitation. In our case, I'll advise you to manually create 2 extra rows for the first and last date with 0 as value for each series and select "no extrapolation" (but keep interpolation) to be sure they are all interpolated between these 2 dates only. 

0 Kudos
EdBerth
Level 3
Author

thanks for the advice

I understand that interpolation is not possible but I thought extrapolation worked.

 

Can you help me create these extra rows for each customer ?

0 Kudos
EdBerth
Level 3
Author

I do it with a python recipe

that works

thanks

 

0 Kudos