Feature Extraction Techniques to Analyze Competitive Exam Result Data to Improvise Educational Policies
Richa1, Jyoti Singh2
1richa12.nitrr@gmail.com, Hewlett Packard Enterprise, STSD, Bangalore, Karnataka, 560016, India
2jsraipur13@gmail.com, Chhattisgarh Professional Examination Board, Raipur, Chhattisgarh, 492002, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The increase in e-learning resources and online facilities for applications, examinations and evaluations has generated huge amount of data in educational field. As quality of education plays vital role in development of individuals, this work aims at analyzing the quantity as well as quality of education district wise, in a state, by utilizing a competitive exam result database. In some district, the results may be better quantitatively, while in others, number of students cleared may be less, but scores may be higher. The aim is to understand the strength and weakness of existing educational infrastructure and how to bring up its level in rural districts, thus helping policymakers to reform and improvise. The collected database undergoes pre-processing phase to ensure its quality and feature engineering is done for missing values, data validation, inconsistencies, data dependencies and outliers. Data inconsistency is handled to ensure data integrity and reliability. Outliers are detected and removed to avoid skewing the predictions on examination results. Machine learning and deep learning models will be applied to Exam-result database to analyse and extract valuable insights. This knowledge can aid in optimizing resource allocation, improving policies to enhance overall educational outcomes and implementing targeted interventions. The accurate predictions generated from this work can help to identify applicants’ risk of underperforming and can provide personalized support
Key Words: Feature extraction, Information retrieval, Classification, Data pre-processing, Data analysis, Education assessment, Correlation analysis, Skewness, Kurtosis, KNN algorithm, Gaussian Naïve Bayes algorithm, Bernoulli Naïve Bayes algorithm