Classification of non-proliferative diabetic retinopathy subjects and healthy subjects with an improved accuracy rate using random forest and random tree classifiers

In order to differentiate between healthy persons and those who have non-proliferative diabetic retinopathy, the study that is proposed to be conducted would make use of the most advanced Random forest and Random tree classifier possible. Detailed instructions and materials: During the course of our proposed study, we made use of the Messidor dataset, which had 138 individuals who were categorised as abnormal and 138 participants who were categorised as normal. For the purpose of this study, we kept the alpha error at 0.5, the threshold at 0.05, the confidence interval at 95 percent, and the G power at 80 percent. Within the scope of our proposed research, we utilised the Python programming language to implement the Principal Component Analysis (PCA) feature selection methodology for the purpose of feature selection. Using cutting-edge machine learning methods such as Random Forest (RF) and Random Tree (RT), the people were classified as either normal or aberrant through the use of these approaches. For the purpose of carrying out the statistical analysis, IBM’s SPSS software, version 22, was utilised. The results of the study revealed that there was a statistically significant difference between two groups (p<0.05), with a p-value of 0.000. When compared to the Random Tree Classifier, the Random Forest Classifier achieved a higher level of classification accuracy, achieving 96.02 percent overall (94.11 percent).