Using Boosted k-Nearest Neighbour Algorithm for Numerical Forecasting of Dangerous Convective Phenomena

The authors propose a boosted k-nearest neighbour algorithm for numerical forecasting of dangerous convective phenomena. The algorithm is applied for processing the output data of the numerical cloud model. The results show that boosted algorithm is able to predict the very fact of convective phenomena occurrence with the high accuracy, but it is not so good in distinguishing the specific type of convective phenomena. Comparison with the k-NN algorithm without boosting shows that boosting resulted in better accuracy of forecasting.