Machine learning for differentiating heat-sensitive and tolerant lines in Arabidopsis using high-throughput phenotypic data

+70 phenotypic traits obtained from high-throughput phenotypic data of control, 3 h, 6 h and 9 h heat treated WT and hsp101 Arabidopsis rosette.

The aim is to determine Col-0 and hsp101 plants based on the phenotypic dataset and ranked the most indicative parameters for classification. Logistic regression was used for modeling, and the accuracies were compared using a subset of traits (top 10 traits) versus using all the traits, as well as accuracies for different treatment groups.