search

prio-data / views_transformation_library

The data transformation library used by the views data transformer service
Other
1 stars 1 forks source link

Imputation Routines #3

Open Peder2911 opened 3 years ago

Peder2911 commented 3 years ago

Interpolation of missing data is a very important feature to implement. My suggestion is that we start out with a simple interpolation mechanism, and then open up to more advanced implementations as alternatives.

@doktorandahl will be a key contributor to future more advanced kinds of interpolation.

An interpolation is defined as taking a series:

0, NaN, 1, 0 ,0

And replacing the NaN(s) with meaningful values to allow us to use the whole series.

I might implement a very naïve interpolator as a proof of concept that we can expand upon.

Doktorandahl commented 3 years ago

Thanks, I agree that this is super important. As soon as I have cleared out a few papers on my desk, I will start the process of evaluating imputation routines.

Peder2911 commented 3 years ago

Awesome!