HYBRID REDUCTION ALGORITHM WITH CAT SWARM OPTIMIZATION FOR CHURN PREDICTION

Customer complexity is a main issue and for large companies is the main problem. Considering the immediate impact on firms ’earnings, companies are trying to change strategies to calculate customer concerns. Consequently, it is very important to find a way to solve this problem by differentiating the factors that increase the client's depression. The chief involvement of this study is to progress an effective churn prediction prototypical using a hybrid approach. Here, initially, data is collected from the dataset and the missing data is removed
at the pre-processing stage. Then, to reduce the problem, the input dataset is enhanced as a dimension reduction function. For
dimensional reduction, the proposed method uses a hybrid technique. Here, PCA and LDA algorithm are hybridized to reduce
dimensionality. After the dimensionality reduction process, the reduced dataset is provided to the optimal continuous neural network
(ORNN). Here, the traditional RNA classifier is trained with Cat Swarm Optimization (CSO). In this work, Tera Data Center at Duke University churn set of predictive data for the calculation, the measured performance. Finally, the performance of the proposed model is estimated at different scales, and it is recognized that the proposed system, designed with dimensional reduction through optimal
classification methods, performs better with 95.08% classification accuracy compared to other classification models