featuretools
featuretools copied to clipboard
An open source python library for automated feature engineering
Created a separate issue to discuss / track idea brought up in #203 by @candalfigomoro more info on Koalas: https://github.com/databricks/koalas
Currently users must supply a dictionary of variable types when creating and entity from a Dask dataframe as inferring variable types from a Dask dataframe is not currently supported. This...
After we release featuretools with support for dask dataframes, we should add a reference to that in the dask tutorial currently up here: https://community.alteryx.com/t5/Data-Science-Blog/Scaling-Featuretools-with-Dask/ba-p/545687
Update `ft.encode_features()` to support encoding a Dask feature matrix, returning a new encoded Dask feature matrix. `encode_features()` currently only works when passing in the feature matrix as a pandas dataframe.
In #881 we tested to see how the new general purpose dask dataframe implementation compared to our existing dask bag approach. We got similar enough numbers and were happy. Next,...
Related to supporting Dask and Koalas DataFrames, we should support cuDF DataFrames to enable us to take advantage of GPU acceleration.
In `calculate_feature_matrix`, there is a call that converts a dataframe column into a list of instance ids using `list(df[index_var])` if cutoff times are not supplied as a dataframe. This is...