r - Appropriate transformation for model that is overestimating on low probabilities - Cross Validated


i have model overestimating low probabilities per chart below. know if underfitting there apply sigmoidal transformation. current model have appears instable , whilst doubles prediction rate (of low frequency event) reluctant use in current state.

the chart comparing event probabilities against binned predicted probabilities.

i've posted here r question stats question.

any thoughts? enter image description here j


Comments

Popular posts from this blog

angular - Ionic slides - dynamically add slides before and after -

minify - Minimizing css files -

Add a dynamic header in angular 2 http provider -