Enhancing Classification and Prediction through the Application of Hybrid Machine Learning Models

Abstract

This paper aims to improve the computational prediction efforts of machine learning models in a health industry project by applying combined models. As the data universe expands exponentially, so do the need and the expectations for machine learning models' prediction accuracy and trustworthiness. Hybrid machine learning models were created, trained, validated, tested, and demonstrated the expected prediction outcomes to address better prediction efforts. We used five machine learning models to develop and assess two hybrid models: logistic regression, random forest, k-nearest neighbour, decision tree, and XGBoost. We discovered that merging machine learning models enhanced the prediction effort observed in the hybrid models' malignancy prediction outcomes. The proposed hybrid models were suitable for predicting classification challenges such as the Wisconsin breast cancer dataset. Improved prediction findings can assist medical organisations in making more accurate diagnoses and informed decisions. Based on the observed prediction speed and outcomes, the hybrid models applied were appropriate for distinguishing patterns of non-cancerous data samples from cancerous ones.