Survey on Data Mining and Predictive Analytics Techniques

Abstract

Nowadays, predictive analytics is one of the most important big data trends. Predictive analytics is the accumulation of extensive, mostly unstructured data from various sources. The mixture of various information sources, for example, online networking information, climate and traffic are improved by internal information is especially basic. But both predictive analysis and data mining attempt to make divination about possible events in the future with the help of data models. Predictive analytics processes utilize various statistical strategies such as machine learning or neural networks, regression and extrapolation to perceive in the information patterns and infer algorithm. These algorithmic procedures are assessed depending on test data and optimized data. It is to be noted that as data availability increases, the accuracy of the algorithm also improved. By chance if the improvement procedure is finished, the algorithm and the model can be connected to information whose classification is obscure. Predictive analytics model captures connection between various factors to assess chance with a specific set of conditions to distribute a score or weightage. Successfully, on applying predictive examination, the organizations can adequately explain huge information for their benefit. We present a detailed survey on data mining and predictive analytics here, by analyzing 15 techniques from standard publishers (IEEE, Elsevier, Springer, etc.) of the year from 2008 to 2018. Based on the algorithms and methods utilized which are inconvenient, the problems are analyzed and classified. Moreover, to indicate the improvement and accuracy of all the research articles is also discussed. Furthermore, the analysis is carried to find the essential for their approaches so that we can develop a new technique to previse the future data. Eventually, some of the research issues are also inscribed to precede further research on the similar direction.