AMS_data: AMS data object class
In ant-stephenson/lhc: LHC Higgs Boson Classification Challenge

Description Fields Methods

This reference class object is used to store the AMS metric of a classification model at different decision thresholds. AMS is a performance measure which includes the sample weightings and is defined by the Higgs Boson Kaggle Competition. A vector of true sample classifications (0 or 1), a vector of estimated probabilities from a model, and a vector of scaled sample weights are needed to initialise.

y: A vector of true sample classifications (0 or 1),
prob: A vector of the samples estimated probabilities from a model
weights: A vector of scaled sample weights.
thresholds: A vector of 30 decision thresholds.
ams: A vector of the AMS metric at each threshold.
max_ams: maximum ams
max_thresh: threshold of maximum ams

calc_ams(): Calculate the AMS at each thresholds.
initialize(y, prob, weights): Provide true sample lables, estimated probabilities, and sample weights. A vector of descision thresholds is initalised.
plot_ams(): Plot AMS against threshold.