Fuzzy decision tree based GentleBoost algorithm for detecting chaff echo in weather radar data

Classification is a sort of essential process for extracting meaningful information from real data. There are a lot of popular classification algorithms; GentleBoost algorithm and fuzzy decision tree algorithm are one of them. The GentleBoost algorithm consists of weights and weak classifiers which has slightly better performance than 50%. And the fuzzy decision tree algorithm is a famous classifier that widely used in real applications and has many advantages: robustness, knowledge implementation, and so on. Because the GentleBoost algorithm is a kind of soft-decision classifier with continuous output, the fuzzy decision tree algorithm could be combined as a weak classifier. Therefore, this paper suggests GentleBoost algorithm with fuzzy decision tree for distinguishing chaff echo that has similar features to precipitation echo. The entire process for implementing chaff echo begins clustering methods in space and time domain. With them, training data set is prepared that already separated chaff echo and non-chaff echo by experts and the GentleBoost classifier based on fuzzy decision tree is generated as a result. For verifying the classifier, actual chaff echo appearance case is applied, and confirmed that the classifier can distinguish chaff echo.