Matthew Tovbin

Results 40 issues of Matthew Tovbin

**Problem** Prepping release notes is a boring process. **Solution** Let's add [release-drafter ](https://github.com/release-drafter/release-drafter) to help us do it quicker. **Alternatives** Keep doing it manually which is meh. **Additional context** No...

enhancement

**Problem** Currently `FeatureBuilder.fromDataFrame` only infers a set of primitive TransmogrifAI types directly mapped from dataframe schema, such as `Text`, `Real` etc. But more advanced types, such as `PickList`, `Email`, `Phone`...

enhancement

**Problem** We would like to be able to store some types of metadata of the level of the model, e.g instance holdout splits. Currently there is no easy way to...

enhancement
good first issue

**Problem** We would like to be able to treat GeoLocation values as categorical features. **Solution** Add unary transformer to convert `Geolocation` values into `Country`. **Alternatives** N/A

enhancement
good first issue

**Problem** TransmogrifAI is currently only usable from Scala with Spark. It would be great if one could: 1. Load TransmogrifAI models in Python, display model insights and compare/evaluate with other...

**Problem** Too lazy to merge approved PRs? **Solution** Look no further. Enable Mergify - https://github.com/marketplace/mergify **Alternatives** N/A

enhancement

**Problem** Exported models neither have any version information nor verification checks (on loading) that verify that a particular model can be safely executed with the current code version. **Solution** 1....

**Problem** Some of our transformers & estimators are not thoroughly tested or not tested at all. **Solution** Use `OpTransformerSpec` and `OpEstimatorSpec` base test specs to provide tests for all existing...

enhancement
good first issue

**Problem** Currently there is no standard way for users to test out their readers, features, workflows and apps. **Solution** Similarly to `OpTransformerSpec` and `OpEstimatorSpec` we provide test facilities to allow...

enhancement

**Problem** Currently there is no type checks available to allow response / predictor specific feature engineering. For instance for response feature of type `FeatureLike[RealNN]` it makes sense to have `feature.calibrate`...

enhancement