Insurance Policy Renewable Prediction Model by Machine Learning

Project Discription

this work are dealing with classifying a small or medium sized data. It simply provides good performance accuracy every time. A direct algorithm of C-means method requires time proportional to the product of number of patterns and number of clusters identified. CART algorithm is a decision based algorithm which is used here with C mean algorithm in order to improve the efficiency of classifying data in terms of accuracy ,no of clusters formed to classify the data and the time taken to classify the data from an available vast amount of data. Features of C mean and CART are hybrid here and implemented over the Matlab software tool which generates graphical form of relationships between accuracy, no of clusters identified for classification and the time taken to classify data. This work as an unsupervised classification algorithm while results are better as compared to other existing algorithm. Experiment was done on real dataset where precision values are improved to other existing techniques by 92.7%. It was obtained that proposed work which was combination of FCM_CART so recall value was higher as compared to FCM as well as CART algorithm. This was due to the modification done in CART algorithm. Here input class was identified by the FCM algorithm which improve true positive values of the proposed work. While at the same time F-measure evaluation parameter was also optimize which increase its value by 63.25% against FCM method and 40.4% as compared to CART algorithm.this work are dealing with classifying a small or medium sized data. It simply provides good performance accuracy every time. A direct algorithm of C-means method requires time proportional to the product of number of patterns and number of clusters identified. CART algorithm is a decision based algorithm which is used here with C mean algorithm in order to improve the efficiency of classifying data in terms of accuracy ,no of clusters formed to classify the data and the time taken to classify the data from an available vast amount of data. Features of C mean and CART are hybrid here and implemented over the Matlab software tool which generates graphical form of relationships between accuracy, no of clusters identified for classification and the time taken to classify data. This work as an unsupervised classification algorithm while results are better as compared to other existing algorithm. Experiment was done on real dataset where precision values are improved to other existing techniques by 92.7%. It was obtained that proposed work which was combination of FCM_CART so recall value was higher as compared to FCM as well as CART algorithm. This was due to the modification done in CART algorithm. Here input class was identified by the FCM algorithm which improve true positive values of the proposed work. While at the same time F-measure evaluation parameter was also optimize which increase its value by 63.25% against FCM method and 40.4% as compared to CART algorithm.

Project Sample Image