TB-Case-Prediction

All code and documentation of the project Tuberculosis (TB) remains a significant public health challenge in India, with a considerable burden on healthcare systems and individuals alike. Early detection and intervention are crucial for effectively managing TB and reducing its spread. In this study, we present a machine learning approach to predict the number of future active TB cases in India. We employ two widely used algorithms, multinomial regression and random forest, to develop predictive models based on historical TB data and relevant demographic and healthcare indicators. Our analysis begins with preprocessing the data to ensure quality and relevance. We then compare the performance of the multinomial regression and random forest algorithms in predicting future TB cases. Our results indicate that the multinomial regression algorithm outperforms random forest in terms of accuracy, providing more reliable predictions of future TB cases in India. The predictive model built using multinomial regression leverages various features such as demographic factors, healthcare infrastructure, and historical TB incidence rates to forecast the number of active TB cases. By accurately predicting future TB cases, our model can aid policymakers and healthcare practitioners in allocating resources, planning interventions, and implementing preventive measures effectively. Overall, our study demonstrates the potential of machine learning techniques, particularly multinomial regression, in predicting TB incidence and guiding public health efforts towards combating this disease in India. Further research and refinement of these models could lead to more targeted and impactful interventions, ultimately contributing to the reduction of TB burden in the country. As a result we can say that multinomial regression algorithm gives better accuracy than random forest regression algorithm, in the given set of data. The reason why multinomial regression is better is that, it can handle multi-category outcomes, provides insights into category influences, is flexible regarding variable types, and does not require linearity or homoscedasticity.