Softmax

$P(Y=i|X)=\frac{e^{{\beta }_{0,i}+{\beta }_{1,i}X}}{{\sum }_{j=1}^K{e}^{{\beta }_{0,j}+{\beta }_{1,j}X}}$

• Probability that our prediction $Y$ is of a class $i$ given some data $X$ when there are $K$ classes
• Has the effect of setting one term close to 1 and the others close to 0
• The denominator is a normalization term to make this a valid probability