Performance Evaluation of a Predictive Mode of Student Difficulties, Based on Neural Networks

With the arrival of the internet, huge volumes of data have been generated in several repositories and websites such is the case for social media Twitter, which registers exponential growth due to constant user interactions causing much research focused on several fields of study including the automation of processing such data. Concerning the field of education, there is an increment of studies that combines Big Data with the information derived from such a field of study through the development and implementation of algorithms that utilize different Data Mining techniques, Natural Language Processing, and Machine Learning. The main contribution of this research consists of the development of an algorithm to predict college students’ difficulties, reported on the social media Twitter by using a basic artificial neural network, creating 11 models with different characteristics to identify the model that reaches higher accuracy, and comparing the results with other developed algorithms. The model built on basic artificial neural networks reached 77,78% accuracy, thus, it is considered an acceptable model for the case study.