Introduction to SoftMax Regression (with codes in Python) - DataScribble

Understanding how softmax regression actually works involves a fair bit of Mathematics. It involves concepts like partial differentiation, maximum likelihood function, gradient descent and matrix multiplication. We will begin by understanding how parameters of this regression are found, and how the hypothesis function h(theta) is found to make predictions for new values of the attributes. In the latter part, we will translate our understanding into code and implement it on the famous'iris' dataset for classifying flowers into one of three categories. Interestingly, the sklearn module in Python does not provide any class for softmax regression, unlike it does for linear and logistic regression.