Predictive analysis and the accuracy of patterns in large amounts of data play critical roles in many disciplines. The process of prediction finds important information in a large amount of data that would otherwise be obscured. In the current decade, all business organisations store digital data in the form of databases and cloud storage. The stored data analyses business performance and forecasts future business planning. The planning of business depends on the accurate and precise pattern analysis of big data. Recently, several prediction-based linear and non-linear machine learning and data mining algorithms have been proposed. These algorithms face the problem of multiple features in big data and the declining accuracy of predictive algorithms. This paper proposes a hybrid algorithm based on convolutional neural networks (CNN) and BAT optimization algorithms. The BAT optimization algorithm reduces the variance of multiple features in big data. The optimised features data process of classification uses the CNN model. The proposed algorithm is tested on two standard datasets: KDDCUP2003 and heart disease data. The parametric analysis of the proposed algorithm shows better performance than existing algorithms such as RNN and CNN.

Predictive Analysis, Bigdata, Bat, CNN, RNN, Machine Learning, KDD