Commit History

  1. cleaned up code/print statements/comments

    Katherine
    e703447
  2. ablation fix

    Katherine
    58e5588
  3. Merge branch 'katherine_branch'

    Katherine
    ac5cc9d
  4. revised readme/setup/added diagnose leakage back

    Katherine
    04f77e4
  5. file organization/deleting redundant files

    Katherine
    cfaa360
  6. Merge branch 'main' of https://github.com/sashpol111/burnout-tracker

    Katherine
    17ef969
  7. file organization / deletion

    Katherine
    e3188b7
  8. rewrite comments in data_loader.py and data_pipeline.py

    fd99b8f
  9. created data pipeline engineering: 3 sources (apis and synthetic data) merged into 1 data set

    Katherine
    3239ea2

  1. fixed error analysis

    Katherine
    05977bf
  2. rewrote / fixed error_analysis. generates png of 4 visualizations

    Katherine
    8799217
  3. fixed ai coach--multi-step ML pipeline that connects multiple models

    Katherine
    04fe416
  4. implemented regularization techniques: l1/l2 & early stopping

    Katherine
    9b132e1
  5. fixed deterministic issue + created preprocessing data pipeline

    Katherine
    4e48d40
  6. preprocessing

    Katherine
    003c0b8

  1. DistilBERT fine-tuning complete - F1 0.753 AUC 0.925

    0753fe5
  2. synthetic data pipeline - 310 entries generated, F1 improves 0.893 to 0.895

    b025afa
  3. add synthetic data generation, error analysis, train with synthetic pipeline

    be84f42
  4. add error analysis with confusion matrix and failure case analysis

    1ed515e
  5. ablation study - 5 feature set experiments documented

    4f56e6d
  6. fix system_prompt bug in get_burnout_advice

    768b66b
  7. add Groq LLM advisor with burnout coaching

    c6e3694
  8. add visualizations - feature importance, training curves, model comparison

    12275db
  9. tune XGBoost with regularization - train 0.971 val 0.954

    667d4b2
  10. add data pipeline, baseline, and XGBoost model

    c97e168