The target variable is continuousRegressionsLinear RegressionLogistic Regression Regression NotionCost FunctionIRLSMulti-Class RegressionExpectile RegressionQuantile Regression Mean Squared Error < Cross Entropy It's not that regression is bad, but that classification is better as the final evaluation loss function, according to Google Deepmindarxiv.orghttps://arxiv.org/pdf/2403.03950.pdf