Abstract:
One-hot encoding of categorical variables is often needed in statistical models. High-dimensional feature vectors make this strategy fail as categories increase. One-hot encoding also lacks morphological information for string entries. High-cardinality string categorical variables need low-dimensional encoding. These should be scalable to many categories, user-interpretable, and statistically useful. We present a Gamma-Poisson matrix factorization on substring counts and a min-hash encoder for fast string similarities. Min-hash simplifies set inclusions into inequality relations. Both methods scale and stream. These methods improve high-cardinality categorical supervised learning. If scalability is key, the min-hash encoder is best because it does not require data fit; if interpretability is key, the Gamma-Poisson factorization is best because it can be interpreted as one-hot encoding on inferred categories with informative feature names. Both models enable string autoML without feature engineering or data cleaning.
Note: Please discuss with our team before submitting this abstract to the college. This Abstract or Synopsis varies based on student project requirements.
Did you like this final year project?
To download this project Code with thesis report and project training... Click Here
