Employee Salary Prediction

This project predicts whether an employee earns more than $50K per year using demographic and work-related features. It demonstrates a complete machine learning workflow, from data exploration to deployment via a Streamlit web app.

Table of Contents

• Overview
• Features
• Project Structure
• Setup Instructions
• Usage
• Model Details
• Streamlit App
• License

Overview

The goal is to classify employees into two salary classes (>50K or <=50K) based on features such as age, education, occupation, and more. The workflow includes data cleaning, preprocessing, feature engineering, model training, evaluation, and deployment.

Features

• Data cleaning and preprocessing
• Feature selection using Random Forest importance
• Model training (Logistic Regression, Decision Tree, Random Forest, Gradient Boosting)
• Model evaluation and comparison
• Saving the best model pipeline
• Interactive Streamlit app for single and batch predictions

Project Structure

.
├── Employee_Salary_Prediction.ipynb   # Main notebook with ML workflow
├── app.py                             # Streamlit app for predictions
├── salary_pipeline_streamlit.pkl      # Saved ML pipeline for deployment
├── adult_3.csv                        # Dataset (not included here)
└── README.md                          # Project documentation

Setup Instructions

1. Clone the repository

   git clone <repo-url>
   cd "Employee Salary Prediction"
   

2. Install dependencies
   It is recommended to use a virtual environment.

   pip install -r requirements.txt
   

   Or manually install:

   pip install pandas numpy scikit-learn streamlit joblib matplotlib
   

Usage

Running the Streamlit App

To run the Streamlit app, execute the following command in your terminal:

streamlit run app.py

This will launch the app in your default web browser. You can enter employee details in the sidebar to get a salary class prediction, or upload a CSV for batch predictions.

Batch Prediction

To predict salary classes for multiple employees at once:

1. Prepare a CSV file with the following columns (matching the training features):
   • age, workclass, education, occupation, hours-per-week, educational-num
2. Use the "Batch Prediction" section in the app to upload your CSV.
3. Download the results with predicted classes.

Model Details

• Features Used:
  • Only the most important features (selected via Random Forest feature importance):
    • age, workclass, education, occupation, hours-per-week, educational-num
• Preprocessing:
  • Handles missing values, encodes categorical variables, and scales numerical features
• Model Selection:
  • Compared Logistic Regression, Decision Tree, Random Forest, Gradient Boosting
  • Selected the best model (Gradient Boosting) based on accuracy and F1-score
• Deployment:
  • The final pipeline (preprocessing + scaler + model) is saved as salary_pipeline_streamlit.pkl
• Model Performance:
  • The model achieved an accuracy of 85.5% and an F1-score of 0.67.

Streamlit App

The Streamlit app provides:

• Single Prediction:
  • Enter employee details in the sidebar and click "Predict Salary Class" for an instant prediction.
  • 
• Batch Prediction:
  • Upload a CSV file (with the required columns in the correct order) for batch predictions.
  • Download the results as a CSV with predicted classes.
  • 
• User-Friendly UI:
  • Clean, interactive interface for both individual and batch use cases.
  • Preview of uploaded data and predictions.

Note: The app requires salary_pipeline_streamlit.pkl to be present in the project directory.

License

This project is licensed under the MIT License.

---

Project completed under IBM internship program taught by Edunet Foundation through AICTE.