How rahulrayy/can-llms-make-tabular-ml-better works
This repository is a notebook-based experiment comparing raw-feature, LLM-feature, and combined tabular classifiers. Its workflow samples and preprocesses Adult Income and HMEQ data, generates cached LLM descriptions, converts those descriptions into structured and embedding features, then evaluates XGBoost variants with nested cross-validation and statistical tests. The README also identifies PCA sensitivity analysis, SHAP analysis, and prompt ablation as experiment capabilities. Only the README is available as evidence; the notebook’s internal implementation and exact call structure could not be verified, so the graph shows documented relationships only.
Main entry points: Can_LLMs_Make_Tabular_ML_Better.ipynb Nested cross-validation
Components 18
Data preparation
- Adult and HMEQ data
Can_LLMs_Make_Tabular_ML_Better.ipynb - Stratified sampling
Can_LLMs_Make_Tabular_ML_Better.ipynb - Categorical encoding
Can_LLMs_Make_Tabular_ML_Better.ipynb
LLM enrichment
- Prompted feature generation
Can_LLMs_Make_Tabular_ML_Better.ipynb - Groq LLaMA API
- Google Drive cache
- Structured output parsing
Can_LLMs_Make_Tabular_ML_Better.ipynb - Text embeddings
Can_LLMs_Make_Tabular_ML_Better.ipynb - MiniLM encoder
Model evaluation
- Feature matrix variants
Can_LLMs_Make_Tabular_ML_Better.ipynb - Nested cross-validation
Can_LLMs_Make_Tabular_ML_Better.ipynb - Fold-local PCA
Can_LLMs_Make_Tabular_ML_Better.ipynb - Optuna tuning
Can_LLMs_Make_Tabular_ML_Better.ipynb - XGBoost classifiers
Can_LLMs_Make_Tabular_ML_Better.ipynb
Experiment analysis
- Statistical comparison
Can_LLMs_Make_Tabular_ML_Better.ipynb - SHAP analysis
Can_LLMs_Make_Tabular_ML_Better.ipynb - Prompt ablation
Can_LLMs_Make_Tabular_ML_Better.ipynb
Other components
- Researcher
Connections 20
- Researcher → Stratified sampling: runs experiment
README.md - Adult and HMEQ data → Stratified sampling: samples rows
README.md - Stratified sampling → Categorical encoding: prepares rows
README.md - Categorical encoding → Prompted feature generation: formats readable rows
README.md - Prompted feature generation → Groq LLaMA API: sends prompts
README.md - Prompted feature generation → Google Drive cache: caches responses
README.md - Groq LLaMA API → Prompted feature generation: returns descriptions
README.md - Prompted feature generation → Structured output parsing: parses output
README.md - Structured output parsing → Text embeddings: encodes notes
README.md - Text embeddings → MiniLM encoder: encodes text
README.md - Categorical encoding → Feature matrix variants: provides raw features
README.md - Structured output parsing → Feature matrix variants: provides structured features
README.md - Text embeddings → Feature matrix variants: provides vectors
README.md - Feature matrix variants → Nested cross-validation: evaluates variants
README.md - Nested cross-validation → Fold-local PCA: fits within folds
README.md - Nested cross-validation → Optuna tuning: tunes parameters
README.md - Nested cross-validation → XGBoost classifiers: evaluates models
README.md - Nested cross-validation → Statistical comparison: compares scores
README.md - Nested cross-validation → SHAP analysis: supports analysis
- Prompted feature generation → Prompt ablation: varies prompt framing