Resume-Analyser

Local, privacy-friendly resume analysis and classification using traditional ML and modern embeddings. Convert resumes, train a classifier, predict categories, and generate actionable advice — all on your machine.

Features

• Resume-to-text conversion for CSV and PDF datasets
• TF‑IDF + Logistic Regression baseline classifier
• Optional sentence-transformer embeddings for richer features
• Simple CLI scripts for converting, training, predicting, and advice
• Configurable via config.yaml

Quickstart

1. Create and activate a virtual environment (recommended).

python -m venv .venv
. .venv/Scripts/activate  # Windows PowerShell: .venv\Scripts\Activate.ps1

2. Install dependencies.

pip install -r requirements.txt
# Temporary compatibility pin
pip install numpy==1.26.0 --force-reinstall

3. Prepare data (see Data section) and run the pipeline below.

Data

• Training data: Kaggle: UpdatedResumeDataSet.csv
• Test data: Kaggle: Resume Dataset

Recommended layout:

data/
  raw/
    UpdatedResumeDataSet.csv
  test/
    <your_pdf_files>.pdf
  processed/
    converted/

Usage

1) Convert resume data to plain text

CSV to text:

python src/convert_dataset.py \
  --csv data/raw/UpdatedResumeDataSet.csv \
  --outdir data/processed/converted

PDF directory to text:

python src/convert_test_data.py \
  --pdfdir data/test \
  --outdir data/processed/converted_test

The CLI naming may be refined in future iterations.

2) Train the classifier

python src/train_classifier.py

Example output plot:

Additional algorithms will be experimented with in future updates.

3) Predict a category for a processed resume

python src/predict.py --input <path_to_text_file>

4) Generate resume advice (MVP)

python src/advice.py --input <path_to_text_file>

The current advice checks for:

1. Length — is the resume too short?
2. Missing keywords — e.g., "Python", "machine learning"
3. Missing sections — Experience, Projects, Education, Skills
4. Soft skills — mentions of communication, leadership, etc.
5. Role match — proximity to a target career path (e.g., Data Science)

This is an early MVP and will evolve.

Configuration

Tune behavior via config.yaml:

model:
  embedding: all-MiniLM-L6-v2
  tfidf_max_features: 1000
  advice_threshold: 0.5

Project Structure

src/
  convert_dataset.py        # CSV → text conversion
  convert_test_data.py      # PDF dir → text conversion
  train_classifier.py       # Train baseline classifier
  predict.py                # Predict class for a resume text
  advice.py                 # Generate heuristic advice (MVP)
models/                     # Saved models/artifacts
data/                       # Raw/test/processed data
assets/                     # Plots and images

Roadmap

• Experiment with additional classifiers (SVM, RandomForest, XGBoost)
• Improve advice heuristics and scoring
• Add evaluation on held-out test sets and reporting
• Streamline CLI and naming for consistency
• Optional lightweight web UI

Contributing

Contributions, issues, and feature requests are welcome! Feel free to open a PR or issue.