Automated Student Performance Analyser And Recommender

Abstract

Big data is a torrent of data streams that are complex such that traditional data-processing system software is inadequate to deal with this huge data. Dealing with big data consists of challenges such as capturing data, data processing, data storage, data analysis, search, sharing, transfer, visualization, querying and updating. Predictive modelling is used for analysing historical events or data to predict known or unknown facts. Neural networks is used to train models efficiently based on the input data sets consisting of parametric features provided. Educational institutions deal with voluminous student mark records that are utilized for understanding and interpreting the institutional academic competency with other institutions. Analysing the performance involves manual computations for report generation. In this paper we have introduced a student performance analyser and recommender that uses prediction algorithms and content based recommendation approach to predict the academic performances which is fully automated and reduces the manual calculations while enabling students to select from a range of recommended subjects based on multiple queries that are suitable to the caliber of the student. The prediction algorithm uses back propagation techniques that take multiple input parametric features to improve the performance in terms of reducing the error rate. Multiple parametric feature
inputs are integrated and compared to the performance when single input features are provided. This research paper helps in analysing the data by automating the tedious calculations including prediction of students performance and recommending papers for the upcoming year. It has been observed that based on the analysis reports that are automatically generated with prediction, future performances of students are found to be accurate due to the multiple input features that gave a higher accuracy and low error rate when compared to other traditional models.