CODEWITHSUNDEEP
pythonmissing-datamultiple-mice
Sjoerd Braaksma
Sjoerd Braaksma

Reputation: 76

Saving a miceforest kernel

According to the miceforest documentation (https://github.com/AnotherSamWilson/miceforest#Imputing-New-Data-with-Existing-Models) it is stated to be possible to save kernels and re-use them later to impute missing data.

However, I can't find any command to actually save the kernel to disk & load the kernel for later usage. Is it possible to save the kernel object somewhere? Or do i need to store the underlying models used for imputation separately? The documentation only shows the kernel being used on a different dataset.

Thanks in advance!

Upvotes: 0

Views: 427

Answers (1)

Suspicious_Gardener
Suspicious_Gardener

Reputation: 126

There is now a kernel.save_model() method, which converts the working data to a parquet and compresses the rest of the kernel. The kernel can then be loaded using the miceforest.utils.load_kernel() function.

Upvotes: 1

Related Questions