Predicting Blood Donations with Regression Algorithms

Abstract

Blood transfusion is a critical procedure that saves countless lives by replacing lost blood during major surgeries, injuries, and treatments for various illnesses and blood disorders. However, maintaining an adequate blood supply remains a significant challenge for healthcare providers. This study focuses on predicting the likelihood of a donor returning to donate blood in the future, utilizing data collected. Our approach includes rigorous data pre-processing, such as feature engineering, normalization, and stratified data splitting. The study further explores logistic regression resulting in a marginally improved AUC score of 0.7890. The findings demonstrate the potential of machine learning models to predict donor behaviour, offering insights that can enhance blood donation campaigns and resource allocation. The study emphasizes the importance of accurate predictions in blood donation forecasting, particularly during periods of fluctuating demand, such as holiday seasons. Even minor improvements in model accuracy can significantly impact the effectiveness of donor recruitment strategies.