This report uses MISR data to model cloud detection. Signals recorded by MISR are used to predict whether something in an image is or is not a cloud. After initial EDA, we develop several classifiers, asses their fit using a suite of metrics, and finally choose the best model. For this study, we considered logistic regression, random forest, quadratic discriminant analysis and naive bayes.
While we have not published our dataset, similar data can be obtained from the NASA Langley Research Center Atmospheric Science Data Center.