Analysis and Predictive Modeling of the Total Literacy Rates in India for Different States

Ayush Yajnik, Vishal Sharma

Abstract: Objectives: To analyze and predict literacy rates across 36 states and union territories in India, focusing on 478,000 primary, upper primary, and secondary schools using Educational Data Mining techniques. Method: This study employs the CART (Classification and Regression Trees) Machine Learning Algorithm to develop a predictive model for literacy rates. Data from 2015-2016 academic year, covering 27 variables related to school infrastructure and demographics, was collected from the Government of India's official education database. The dataset was preprocessed, normalized, and analyzed using Python's pandas and matplotlib libraries. Findings: The CART algorithm successfully identified significant predictors of literacy rates with 79% accuracy. Key findings include: (1) A 24% chance of literacy rates around 58% in states where primary schools accessible in all weather conditions exceed 1,951. (2) A 39% probability of 56% literacy rate in states with less than 438 primary schools with electricity. (3) Rural areas showed 15-20% lower literacy rates compared to urban areas. (4) States with over 436 primary schools with playground facilities had a 3% chance of achieving a 73% literacy rate. Novelty: This research contributes to Educational Data Mining by providing the first comprehensive analysis of literacy rates across all Indian states using the CART decision tree algorithm. It quantifies the impact of specific infrastructural factors on literacy, emphasizing the need for data-driven strategies in addressing educational disparities.

Keywords: Data Mining, CART Machine Learning Algorithm, Literacy Rate Prediction, Educational Data Mining