A java implementation for calculating the accuracy metrics (Accuracy, Error Rate, Precision(micro/macro), Recall(micro/macro), Fscore(micro/macro)) for classification tasks based on the paper [A systematic analysis of performance measures for classification tasks] (http://www.sciencedirect.com/science/article/pii/S0306457309000259) and MATLAB confusion implementation.
public static void main(String[] args) {
// targets: SxQ (S:Classes; Q:Samples)
// outputs: SxQ (S:Classes; Q:Samples)
double[][] targets = {
{1, 1, 0, 0, 0, 0},
{0, 0, 1, 1, 0, 0},
{0, 0, 0, 0, 1, 1}
};
double[][] outputs = {
{0.1, 0.86, 0.2, 0.1, .02, 0.1},
{0.4, 0.12, 0.768, 0.145, 0.1, 0.8},
{0.454, 0.35, 0.21, 0.0, 0.89, 0.9999}
};
Confusion confusion = new Confusion(targets, outputs);
confusion.print();
Evaluation evaluation = new Evaluation(confusion);
evaluation.print();
}Confusion Results
Confusion value
c = 0.17
Confusion Matrix
1 0 1
0 2 0
0 0 2
Indices
[1] [] [0]
[] [2,3] []
[] [] [4,5]
Percentages
0.2 0.0 1.0 0.8
0.0 0.0 1.0 1.0
0.0 0.33 0.67 1.0
Accuracy Evaluation Results
Average Accuracy(%) : 91.11
Error(%) : 8.89
Precision (Micro)(%) : 88.89
Recall (Micro)(%) : 93.02
Fscore (Micro)(%) : 90.91
Precision (Macro)(%) : 88.89
Recall (Macro)(%) : 94.44
Fscore (Macro)(%) : 91.58
For C++ Implementation, visit accuracy-evaluation-cpp
For MATLAB Implementation, visit accuracy-evaluation-matlab