ISCHEMIC STROKE PREDICTIVE ANALYTICS USING FUSION GRADIENT BOOSTING DECISION TREE ALGORITHM COMBINING LIGHTGBM AND XGBOOST IN BIGDATA
Health care is a wide area where data plays a major role. Scope of Big data analytics is extremely high and important for the prediction and prevention of various diseases. The Big Data analytics is effectively applied since we have extremely diversified data form from different sources. In this paper we have proposed a fusion prediction model for Ischemic Stroke based on the Light GBM algorithm and the XGBoost algorithm with GBDT. The prediction model is built using the user's physical examination data and major biomedical indicator like blood pressure, BMI, Heart diseases, Avg glucose level, and Cigarette smoking are used as the auxiliary judgment criteria. The LightGBM model's predicted results and the XGBoost model's expected results are then both fed into the GBDT model. Finally, the fusion of the two prediction results is obtained. This fusion prediction model results are compared with various other machine learning classifiers. The experimental result proves that the fusion predictive model is more accurate than the other classifiers.
Big Data Analytics, Light GBM, XG Boost, GBDT