Abstract: The authors analyzed the model of harmful algal blooms in the river on the basis of classification regression tree (CART) algorithm of data mining. Results indicated that phosphate, chloride and the maximum pH values are key factors of algae generation. Furthermore, we employed the R language to validate the superiority and convenience of using CART algorithm. The conclusions and methods could contribute to a more effective water quality monitoring and forecasting.

Key words: data mining, classification and regression tree (CART), R language, water quality monitoring