ML Placement Predictor
Built a complete machine learning pipeline from data preprocessing to deployment. The model uses Logistic Regression trained on student academic data to predict campus placement outcomes. Features include interactive data visualization and real-time prediction through a Streamlit web interface.
1. Overview
This Python machine learning project is an end-to-end web application that predicts campus placement outcomes. By leveraging a Scikit-learn logistic regression project pipeline and deploying it via a Streamlit machine learning app, it provides real-time placement prediction machine learning capabilities.
2. Problem Statement
Colleges and students often lack early indicators for campus placement success. This student placement ML project aims to predict whether a college student will get placed based on early academic and cognitive indicators (CGPA and IQ) to assist in early intervention.
3. Dataset
A structured dataset containing student academic metrics, specifically their CGPA and IQ scores, mapped against their final campus placement outcomes.
4. Methodology & Data Preprocessing
The machine learning prediction project pipeline handles feature variance using a serialized StandardScaler configuration to normalize raw metrics. This ensures live user inputs match the exact mathematical distribution of the training data.
5. Analysis & Model
A Logistic Regression classifier was trained and evaluated to optimize the decision boundary separating placed and unplaced student data clusters.
6. Results & Key Findings
The student placement prediction machine learning model achieved an 84.7% accuracy in predicting placement outcomes, successfully deployed in a lightweight, responsive dashboard.
7. Limitations
The current model relies heavily on CGPA and IQ, without accounting for soft skills, interview performance, or extracurricular activities which heavily influence actual placements.
8. What I Learned
Gained practical experience in deploying a Scikit-learn project into production using Streamlit, and managing feature scaling states between training and inference environments.