Interview Practice Questions

When can single decision tree outperform random forest?

How does one-hot encoding work for categorical features?

What to do with rare categories in train/test split?

How to use SHAP to analyze feature influence on specific prediction?

How to process features containing click-through data?

How is feature importance calculated in different models?

How to properly link time series together for analysis?

How to document and version ML experiments?

How to compare ML models objectively?

How do threshold algorithms work with windowed summation?

How to select last N records in temporal data for analysis?

Problems with non-aggregated ranks in data processing?

How to analyze binary outcomes (sell=1 or sell=2)?

How to handle last elements in sequential data?

How to work with duplicates at the end of datasets?

How to work with naming and grouping by departments in feature engineering?

What does finite number of classes mean in machine learning?

What are precision and recall metrics in machine learning?

How to correctly interpret positive classes in classification?

Difference between predicted and actual in precision/recall context?