Study of Machine Learning Based Rice Breeding Decision Support Methods and Technologies

The Objective of the study is to Analyze and mining rice breeding data with data explore and machine learning algorithms to discover how rice biological characters influence the economic characters, explore effective methods and technologies for breeders and help them find appropriate breeding parents, and provide tools for parental selection in rice breeding. The author developed a B/S application with Python and Django, which implement real-time data mining of rice breeding data. Data analysis and processing result generated from decision tree algorithm can find effective breeding knowledge and patterns, and spectral biclustering algorithm can find required varieties with their local features follow certain patterns. The system can help breeders find useful knowledge and patterns more quickly, and improves the accuracy and efficiency of crop breeding.