3 ms·
ooc do you plan on binding in common functions for the DS/ML use cases? Things like - String similarity measures - ROC-AUC/MSE/correlation/Precison/Recall et
by lumost 6y ago
ooc do you plan on binding in common functions for the DS/ML use cases? Things like
- String similarity measures
- ROC-AUC/MSE/correlation/Precison/Recall etc.
- LSH
- Sampling/joining with random records.
Keeping all of the transformation/prep logic in the sql engine seems like a great performance savings over python, and would also speed up the dev time for building up the code surrounding the ML functionality.
- mytherin 6y agoWe already have a number of statistical ops (e.g. correlation) available, and we are planning to add more. The exact timeline I cannot promise, but feel free to open issues with the specific operations you are interested in/you think will be useful. We are always happy to review PRs as well :)