Skip to content

getml.feature_learning

Feature Learners are getML's powerhouse leveraging the high performance of C++ to ensure efficient execution and effective memory use.

There are five different algorithms, each capitalizing on unique strengths and suitable for different use cases. Most of them are available in the getML Enterprise edition.

FastProp

  • FastProp utilizes aggregation-based operations, enabling rapid generation of numerous features through simple aggregations. This makes FastProp ideal for the exploration phase of a data science project, delivering quick, decent results.
  • FastProp is available in both, the getML Community edition and getML enterprise edition.

Multirel

  • Multirel focuses on minimizing algorithmic redundancies through incremental updates and combining these improvements with ensemble learning methods.
  • Recalculations are only performed where changes occur, significantly increasing efficiency while integrating methods such as bagging and gradient boosting.
  • This feature learner is available in the getML Enterprise edition.

Relboost

  • Relboost extends the gradient boosting approach, to relational learning by focusing on aggregating learnable weights rather than columns, addressing computational complexity and exponentially growing feature space.
  • While Relboost often surpasses Multirel in predictive accuracy and training efficiency, its generated features are less intuitive.
  • This feature learner is available in the getML Enterprise edition.

Fastboost

  • Fastboost uses a simpler, faster, and more scalable algorithm than Relboost, making it ideal for large datasets and many cross-joins. Fastboost can outperform FastProp in speed for datasets with many columns.

  • Fastboost requires free disk space due to extensive memory mapping and has difficulty applying to multiple targets as it must learn separate rules for each.

  • This feature learner is available in the getML Enterprise edition.

RelMT

  • RelMT adapts linear model trees to relational data, combining linear models at each tree leaf to effectively capture both linear and non-linear relationships, making it particularly advantageous for modeling time-series data.
  • This feature learner is available in the getML Enterprise edition.