DATA-DRIVEN ADMISSION PREDICTIONS FOR ENGINEERING COLLEGES USING MACHINE LEARNING

Abstract

Using machine learning (ML), massive amounts of data can be reanalyzed to find patterns that humans may not notice. ML is increasingly used to evaluate educational data like student class performance. In order to improve student performance, data mining, data management, and ML are being used more. Data mining involves extracting hidden data from many raw databases. Thus, knowledge acquisition affects predictive ML models and decision-making. Advanced data mining and ML techniques are now accepted for predicting student exam grades, achievement, etc. Educational data mining involves conventional data mining for educational data analysis to solve educational problems. Artificial intelligence and data mining have improved people's lives. Several million Indian students take the government-run university bachelor's entrance exam. However, only a few thousand pass this competitive exam. Several candidates struggled during this time. However, they were denied admission to an Indian public university, leaving their future uncertain. Family issues, frustration, and admission test anxiety may have prevented their public university admission. The government mostly funds public universities. This project uses machine learning to predict engineering or college admission. Indian students must attend a public university because private universities are too expensive for middle- and low-income families.

Specification

Description:This work implements a machine learning framework for predicting the probability of admission to under graduate program based on certain metrics of a student. In practice, this project provides a systematic approach to loading, preprocessing, analyzing, modeling, and evaluating data for the prediction of admission probabilities in an engineering college. It also includes insights into feature importance and demonstrates how to make predictions using the trained model. Here's an overview of what each section of the code accomplishes:
 Import Libraries: It imports necessary libraries for data manipulation, visualization, and machine learning.
 Data Loading and Preprocessing:
o The dataset "Admission_Predict_Ver1.1.csv" is loaded into a Pandas DataFrame (collegeData).
o Redundant columns are removed (e.g., "Serial No.").
o The "Research" column is converted to a categorical variable since it represents whether a student has research experience or not.
 Data Analysis:
o Correlation between predictor variables and the target variable ("Chance of Admit") is visualized using a heatmap.
o Distribution of "Chance of Admit" based on research experience is visualized using a catplot.
 Train-Test Split:
o The data is split into predictor variables (X) and the target variable (y).
o Further, the data is split into training and testing sets using the train_test_split function.
 Model Building:
o A Random Forest Regressor model is initialized with specific hyperparameters chosen after experimentation.
o The model is fitted to the training data using the fit method.
, C , C , Claims:
1. We claim the framework claims to achieve a high level of accuracy in predicting student admissions based on a comprehensive analysis of historical data and relevant features.
2. We claim by analyzing multiple factors influencing admissions, the framework provides valuable insights that can help engineering colleges refine their admission criteria and processes.
3. We claim the system is designed to be intuitive and accessible, enabling prospective students to easily input their data and receive instant feedback on their admission likelihood.
4. We claim the framework claims to implement advanced techniques to minimize biases in the admission predictions, promoting fair and equitable treatment of all applicants.
5. We claim the architecture of the framework is designed to be scalable, allowing it to handle increasing volumes of data and adapt to evolving admission requirements over time.

Documents