Evaluation of different Machine Learning Models for identifications of Flares with CMEs

Solar eruptions such as CMEs and flares causes geomagnetic and communication disturbances on Earth. CMEs can be found in conjunction with flares, filaments, or independent. Although both flares and CMEs are understood as triggered by a common physical process magnetic reconnection, yet, the degree of association among them is unknown. We attempted to model the association of CMEs with flares through extensive Machine Learning models to study the occurrence of CMEs. Further, to improve the class separability, we have utilized the parameter change information obtained from the respective subsequent time difference. Since there is significant imbalance between the classes, we had explored approaches such as under sampling majority class, oversampling minority class and synthetically generated minority samples through SMOTE Technique. We achieved TSS around 0.81 without adding change information, and TSS around 0.92 after adding change information as additional feature on prediction of CMEs associated with flares for LDA, after addressing the class imbalance issues.