Feature selection¶

The TranzAI platform integrates specialized notebooks dedicated to feature selection. These parameterized notebooks can be instantiated through the TranzAI platform GUI. No code is required. You can directly access to a set of methods (filter methods, wrapper methods) to identify those features that have the strongest relationship with the output variable of your model.

Feature store table time series

TranzAI helps with selecting the best features to use when training machine learning models. This can be helpful in a number of ways:

Models that use fewer features are easier to understand and explain.
It's simpler to create machine learning models when you use fewer features.
Using fewer features can improve the accuracy of your model by preventing it from overfitting (which is when it learns to predict the training data too well and doesn't generalize well to new data).
When you select features, you can remove redundant data which can help improve the accuracy of your model.
Models that use fewer features can be trained more quickly than models that use more features.
Models that use fewer features are less likely to make mistakes.
By selecting features, you can avoid the "curse of dimensionality." Some algorithms don't work well when there are a lot of dimensions (or features) in the data, such as general linear models and decision trees.

Note

You should also consider looking at the stability of selected features in order to increase the success of your model over a long period of time.