A decision tree was developed to aid the understanding of coffee rust epidemics caused by Hemileia vastatrix. Infection rates calculated from monthly assessments of rust incidence were grouped into three classes: reduction or stagnation - TX1; moderate growth (up to 5pp) - TX2; and accelerated growth (above 5pp) - TX3. Meteorological data, expected yield and space between plants were used as explanatory variables for the infection rate classes. The decision tree was trained using 364 examples prepared from data collected in coffee-growing areas between October 1998 and October 2006. The model correctly classified 78% of the training data set and its accuracy was estimated at 73% for the classification of new examples. The success rates of the model were 88%, 57% and 79%, respectively, for the infection rate classes TX1, TX2 and TX3. The most important explanatory variables were mean temperature during leaf wetness periods, expected yield, mean of maximum temperatures during the incubation period and relative air humidity. The decision tree demonstrated its potential as a symbolic and interpretable model. Its model representation identified the existing decision boundaries in the data and the logic underlying them, helping to understand which variables, and interactions between these variables, led to coffee rust epidemics in the field.
Hemileia vastatrix; Coffea arabica; knowledge discovery in databases - KDD; data mining
