Confusion Matrix 3D with Overall PCC and Group Statistics

Displays a 3D confusion matrix, and returns thorough overall stats and stats per group for N groups.

function [confusion_matrix overall_pcc group_stats groups_list] = confusionMatrix3d(predicted_groups,actual_groups)

% DESCRIPTION:
% Confusion matrix-based model performance summary tool.
% Works with character and numeric data, for any number of groups.
%
% Displays your confusion matrix as a 3D bar chart of your observations,
% broken down by their actual and predicted groups.
%
% Assumes that union(predicted and actual_groups) contains all
% possibilities for Groups.
%
% Returns the overall PCC and the following stats per group:
% True Positives, False Positives, True Negatives, False Negatives,
% Sensitivity, Specificity, PCC.
%
%
% OUTPUT:
% 1) a 3D Bar Chart of the number of observations per group predicted as
% each group (helps you visualize the performance of your model in
% predicting each of several groups). X and Y tick labels are the
% names (char or numeric) of your predicted and actual groups in ascending
% alphanumeric order (the same order in the groups_list variable).
%
% 2) confusion_matrix (matrix of doubles): the counts underlying the 3D Bar
% Chart confusion matrix, where columns are different predicted groups, in
% ascending alphanumeric order, and rows are different actual groups, in
% ascending alphanumeric order (the same order in the groups_list variable)
%
% 3) overall_pcc (double): the overall Percent Correctly Classified in your data
%
% 4) group_stats (cell array of structs), where each struct contains:
% group -- the name of the group for the current stat struct
% TP, FP, FN,TN -- True&False Positives&Negatives for the group
% sensitivity -- TP/(TP+FN) for the group
% specificity -- TN/(TN+FP) for the group
% PCC -- (TP+TN)/(TP+TN+FP+FN) for the group
% the cell array's structs are arranged in alphanumeric order of
% group names.
%
% 5) groups_list (cell array of chars or vector): the names of groups in
% alphanumeric order, the same order as they appear on the Confusion Matrix
% 3D Bar Chart and in the group_stats cell array.
%
% INPUTS
% parameter_name (datatype)-- description
% 1) predicted_groups (vector of numeric/logicals, or cell array of chars)--
% The group for each observation, as predicted by your model. If you are
% using a logistic regression model, you need to translate the predicted
% logit scores/ probabilities into groups, based on your own cutoff
% value(s), and then feed those groups into this function.
%
% 2) actual_groups (vector of numeric/logicals, or cell array of chars)--
% The group for each observation, based on your actual data.
%
% Note: if one of these two inputs is a cell array of chars, both need to
% be cell arrays of chars.