Heart Disease Prediction using Innovative Decision tree Technique for increasing the Accuracy compared with Convolutional Neural Networks
Chavana Sateesh, R. Balamanigandan
Abstract
The work aims to determine heart diseases using medical parameters of cardiac patients to improve the accuracy rate to predict in advance. Materials and Methods: One of the Supervised Learning algorithm is Decision tree algorithm, the widely used one for regression as well as classification issues in the problems raised in machine learning. A subset of deep neural networks is CNN, Convolutional neural network. Two algorithms used in this research work performed with five different datasets at each time to record five samples. The sample size is 5 for both the groups using a G power value of 0.8. Results: The decision tree algorithm shows higher accuracy of 87.75% when compared to CNN which has an accuracy of 84.5%. The two algorithms are statistically significant with an independent sample T-Test value is 0.001 (p<0.05) (Sig 2 tailed) with a confidence level of 95%. Conclusion: When comparing two algorithms, random forest significantly seems to be better than CNN.