Resume Classification Web Application

Table of Contents

• Project Description
• Features
• Dependencies
• Installation
• How to Run
• Usage
• Technical Approach
• Model Performance
• Dataset
• Credits
• Contact
• License

Project Description

This project is a web application designed to classify resumes into different job categories using a machine learning model. It helps streamline the resume screening process for recruiters and HR departments by automatically categorizing resumes based on their content.

The application is built using:

• Python: For the backend logic and machine learning model.
• Flask: As the web framework to create the web application.
• HTML/CSS: For the frontend user interface.
• Machine Learning Model: A pre-trained model (Logistic Regression) serialized in a pickle file.

Features

• Resume Upload: Users can upload resumes through a simple web interface.
• Automatic Classification: Resumes are automatically classified into one of 25 job categories.
• Instant Results: Classification results are displayed immediately after the resume is uploaded.
• Text Preprocessing:The app is including:
  • Special character removal
  • Lowercase conversion
  • Stopword removal
  • Tokenization

Dependencies

• Python 3.6+
• Flask
• scikit-learn
• pandas
• nltk
• re
• pickle

Dataset

The dataset used for training the model contains 962 labeled resumes across 25 distinct job categories.

Installation

1. Clone the repository:

Category	Count
Java Developer	84
Testing	70
DevOps Engineer	55
Python Developer	48
Web Designing	45
HR	44
... (20 more categories)	...

Technical Approach

Data Preprocessing

def clean_text(text):
    text = re.sub(r'[^\w\s]', '', text)  # Remove special chars
    text = text.lower()  # Convert to lowercase
    text = ' '.join([word for word in word_tokenize(text) 
                   if word not in stopwords.words('english')])
    return text

Feature Extraction

• TF-IDF Vectorizer with 1000 features
• Text vectorization for machine learning input

Models Evaluated

Random Forest

• Accuracy: 98.96%
• Strengths: Handles non-linear relationships well

Logistic Regression (Selected Model)

• Accuracy: 99.48%
• Strengths: Best overall performance

Multinomial Naive Bayes

• Accuracy: 96.37%
• Strengths: Fast training time

Model Performance

Accuracy Comparison

Model	Accuracy
Logistic Regression	99.48%
Random Forest	98.96%
Multinomial NB	96.37%