Predicting Diabetes Disease using Random Forest Tree (Rft) Data Mining Technique

Diabetes is a condition that happens when the blood glucose is too high, also known as blood sugar. The primary source of energy is blood sugar, and it comes from the food you eat. Insulin, a pancreatic hormone, helps food glucose get into the cells for energy use. It also leads for an unrelated condition named, "Diabetes Insipidus”, which entails complications with the processing of fluids in the kidney. Insulin is the key to the ability of the cell to use glucose. Problems with the processing of insulin or how cells perceive insulin can easily cause out of control the body's carefully balanced glucose metabolism process [1]. Diabetes emerges when either of these conditions happens, blood sugar levels rise and crash and the risk of organ damage. Earlier prediction of this diabetes condition could provide proper treatment to protect the people from un avoided illness. For this prediction we can apply data mining which is used predominantly in healthcare organizations for decision making, disease detection purpose. In this paper data have been collected from UCI repositories and the data mining tool (WEKA) is used to predict diabetes. In this database there are 768 instances in which 500 instances belongs to tested negative and 268 instances belongs to tested positive. An experimental study is carried out using data mining technique classification technique called Random Forest Tree (RFT) classifier to predict diabetes. In this research, we have used different cross fold validation to achieve better accuracy and we found that cross fold validation k= 8 gives high accuracy 76.69% while compared with other cross fold validation values.