Abstract: The occurrence or non-occurrence of diseases is a binary problem. Combined the data of late blight of potato in northern China and meteorological data, the classification model was used to explore the pathogenesis model of potato late blight. Three sample sets were created according to the distance between the weather station and the planting region. Considering the pathogenesis mechanism of late blight of potato, the meteorological data (a day length) and receiver operating characteristic curve (ROC curve) were used in the classification model to select the appropriate parameter threshold and the framework for late blight of potato. Then, the classification model was verified by historical data. The results demonstrated that the overall accuracy of the forecasting model was 78%, the sensitivity was 79%, and the specificity was 78% in the northern planting region.

Key words: medicinal plant, consecutive monoculture problem, barcoded pyrosequencing, bacterial diversity, phyllosphere microorganism