📄 Resume Category Prediction App

A Machine Learning-powered web app that predicts the category of a resume (e.g., Data Science, Network Security Engineer, Advocate, etc.) using NLP and a Support Vector Machine classifier. Built using Python, scikit-learn, and Streamlit.

---

🚀 Features

• ✅ Upload resume files in .pdf, .docx, or .txt formats
• 🔍 Clean and process resume content using NLP
• ✨ Predict job category using a trained ML model (SVM with TF-IDF)
• 📊 Supports over 25 resume categories
• 🌐 Simple, interactive interface using Streamlit

---

📂 Supported Resume Categories

Examples include:

• Data Science
• Network Security Engineer
• Advocate
• Health and Fitness
• Web Designing
• Java Developer
• Python Developer
• Sales
• DevOps Engineer
• HR
• Civil Engineer
• Mechanical Engineer
• Blockchain Developer
  ... and many more

---

🧠 Model Architecture

• Model: Support Vector Classifier (SVC) using One-vs-Rest strategy
• Vectorization: TF-IDF (Term Frequency-Inverse Document Frequency)
• Preprocessing:
  • Remove URLs, mentions, hashtags
  • Remove special characters and punctuations
  • Lowercasing, whitespace removal

---

🛠️ Installation & Setup

1️⃣ Clone the repository

git clone https://github.com/anujayavidmal2002/NLP_ResumeScreeningApp.git
cd NLP_ResumeScreeningApp

2️⃣ Install the required libraries

pip install -r requirements.txt

Or install them manually:

pip install streamlit scikit-learn pandas numpy matplotlib seaborn python-docx PyPDF2

3️⃣ Run the app

streamlit run app.py

---

📁 File Structure

.
├── app.py                    # Main Streamlit app
├── clf.pkl                   # Trained classifier (SVC)
├── tfidf.pkl                 # TF-IDF vectorizer
├── encoder.pkl               # LabelEncoder for category decoding
├── UpdatedResumeDataSet.csv # Original dataset (optional)
├── requirements.txt          # Python dependencies
└── README.md                 # Project documentation

---

📊 Model Training Summary

from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.svm import SVC
from sklearn.multiclass import OneVsRestClassifier
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import LabelEncoder

• Dataset: UpdatedResumeDataSet.csv
• Cleaned resumes
• Encoded with LabelEncoder
• Vectorized using TfidfVectorizer
• Model trained using OneVsRestClassifier(SVC())
• Accuracy: ~90% (on balanced data)

---

🧪 How to Test

You can test the app by uploading a resume in .pdf, .txt, or .docx format.
Example categories include:

• 🧠 Data Scientist
• 🧑‍⚕️ Health and Fitness
• 🔐 Network Security Engineer
• ⚖️ Advocate

---

🧾 Example Resume Snippets

"Experienced Python Developer with background in machine learning, TensorFlow..."
→ Predicted Category: Python Developer

"Certified advocate with 10 years in civil and family law..."
→ Predicted Category: Advocate

---

🧰 Dependencies

• Python 3.7+
• Streamlit
• scikit-learn
• pandas, numpy
• matplotlib, seaborn
• PyPDF2
• python-docx

---

⚠️ Version Compatibility Warning

Scikit-learn model was saved using version 1.7.0.
To avoid version conflicts or InconsistentVersionWarning, make sure to use the same version or retrain the model in your own environment.

---

📜 License

This project is licensed under the MIT License.

---

👨‍💻 Author

Anujaya Vidmal
🖥️ GitHub: @anujayavidmal2002
📧 Email: anujayavidmal2002@gmail.com
🏫 University of Moratuwa

---

📽️ YouTube Demo

🎥 Watch the full demo and tutorial:
🔗 https://youtu.be/F3F6c0N12ls

---

⭐ Acknowledgments

• Dataset by Kaggle community
• Streamlit team for the amazing web app framework