Speaker
Description
The accuracy of machine learning models strongly relies on data quality, including carefully selected features based on established metrics such as Feature importance scores. This paper introduces PermuGini-RF, a hybrid feature selection and classification model that systematically combines Gini Importance for fast upstream screening and Permutation Importance for robust downstream validation within Random Forest architecture for mobile phone price classification. Using the Mobile Price Classification dataset from Kaggle containing twenty hardware features and two thousand mobile phone samples across four price classes, PermuGini-RF was evaluated against single-metric baselines. The dual stage operation of this model adopts Gini Importance for rapid initial feature screening (retaining top 50%), followed by Permutation Importance for final selection and ranking. PermuGini-RF achieves 93.5% cross-validated accuracy, outperforming Gini-only (91.2%) and matching Permutation-only performance with topmost five features while reducing computational cost by 36%. These findings reveal PermuGini-RF as an optimal trade-off model between computational efficiency and selection reliability for mobile phone price prediction tasks.
| Tutorial level (only for Tutorial) | Intermediate |
|---|---|
| Do you plan to submit a 4-page extended abstract on OpenReview (only for Presentations/Posters)? | Yes |