Why is logistic regression not performing well with large sparse features?

one reason I can think of is because of the high dimension large sparse features create… anything else?

Course: Grokking the Machine Learning Interview - Learn Interactively
Lesson: Embeddings - Grokking the Machine Learning Interview

Hi @Roni,

Large sparse features (normally) have more noise and increase the time and space complexity of the model. Therefore, applying logistic regression models to a large space features dataset will fit more coefficients and causes overfitting. Overfitted models are unable to generalize hence they cannot perform well. That is why logistic regression does not perform well with large sparse features.

Do let us know if you still have not gotten the answer.

1 Like

Thanks @Rauf_Tabassam! appreciate the answer!
Also, found this article and wanted to share with the community:

Basically, echoing your answer :slight_smile:

1 Like