Development of a Machine Learning Model For Big Data Analytics

Abstract

The aim of the study was to utilize machine learning approaches in the development of a big data analytics model that can effectively discover patterns and aid in making intelligent decisions. The impact of Big Data on scientific discoveries and technology in Big Data analytics cannot be overstated. In this regard, the study focused on the development of a system to control incoming illegal messages from intruders/unknown users, in order to alleviate illicit contextual communication channels. Unstructured data sets were collected from online blogs and labeled to suit the needs of the machine learning algorithm. The data was preprocessed using Weka machine learning libraries and converted into attribute related file form (arff). The resampling technique was used to partition the data into 80% training set and 20% testing set, which was used to develop the big data model and identify illicit message content to prevent people from being victimized. The Na ï ve Bayes machine learning algorithm was utilized to predict the categorized labels as binary values. The results of the study showed that Naïve Bayes had an accuracy of 96.76% for metrics evaluation of true precision model. It is recommended that the application of other machine learning techniques to the classification model would improve its performance.