Abstract:
As one of the most versatile plants, Jatropha curcas is spread in various regions around the world because of the great benefits it
provides. However, Jatropha curcas is easily attacked by viruses which then cause damage to the plant, such as yellowing leaves
and secreting sap, making it necessary to identify Jatropha curcas disease to deal with the problem as early as possible so that the
losses incurred are not too large. An expert system was built to be able to identify Jatropha curcas disease by adopting expert
knowledge. The use of the Random Forest algorithm as one of the classification algorithms was applied in this study. By using a
random forest, several disease prediction classes are generated by each decision tree that has been formed. The disease class
with the most votes was used as the final result. In this study, the data used were 166 data with 9 diseases and 30 symptoms. The
results showed that Random Forest outperformed other algorithms such as Fuzzy Neural Network and Extreme Learning Machine
with an accuragy of 98.021 using the composition of training data and test data of 70:30.
Keywords: Classification; Decision tree; Disease; Expert system; Jatropha curcas; Random-forest
