Cuckoo-Inspired Algorithms for Selecting Features in the Prediction of Diabetes Using Machine Learning Models

Early diagnosis and prevention of disease are essential. Diabetes is one of primary diseases that cause more human death. According to World Health Organization (WHO) research, it is a significant non-communicable illness that entailed numerous severe health risks. In India illness is spread more widely. The Objective of this work is to employ Machine Learning Techniques to forecast diabetes illness from PIDD dataset. Additionally to enhance performance of model cuckoo technique is used to assess how well Machine Learning Model (MLM) performs. With use of data mining techniques diabetes prediction using PIDD dataset was carried out in this research with a focus on Feature Engineering. MLM is used for predicting after feature engineering. Cuckoo Search (CS) algorithm and Logistic Regression (LR), K -Nearest Neighbor (KNN), Support Vector Classifications (SVC), Gaussian Naïve Bayes (GNB), Decision Tree (DT) and Random Forest (RF) are used in present study. Additionally, validation of best prediction model has been done. Accuracy, precision, recall and F1-Score are evaluation metrics that are taken into consideration for validating models and metrics with highest values are94%, 98%, 94% and 96%. By reducing factors that affect performance of models using cuckoo algorithm accuracy of MLM was boosted.