## Multi-label

Precision/Recall and f-scores all work for multi-label classification, although they have bad qualities in unbalanced classes.

## Unbalanced class problems

đźš§

## Metric Zoo

One of the less abstruse summaries of these is the scikit-learn classifier loss page, which includes both formulae and verbal descriptions; this is surprisingly hard to find on, e.g.Â the documentation for deep learning toolkits, in keeping with the fieldâ€™s general taste for magical black boxes. The Pirates guide to various scores provides an easy introduction.

### Matthews correlation coefficient

Due to Matthews (Matt75). This is the first choice for seamlessly handling multi-label problems, since its behaviour is reasonable for 2 class or multi class, balanced or unbalanced, and itâ€™s computationally cheap. Unless you have a very different importance for your classes, this is a good default.

However, it is not differentiable with respect to classification certainties, so you canâ€™t use it as, e.g., a target in neural nets; Therefore you use surrogate measures which are differentiable and use this to track your progress.

#### 2-class case

Take your \(2 \times 2\). confusion matrix of true positive, false positives etc.

\[ {\text{MCC}}={\frac {TP\times TN-FP\times FN}{{\sqrt {(TP+FP)(TP+FN)(TN+FP)(TN+FN)}}}} \]

\[ |{\text{MCC}}|={\sqrt {{\frac {\chi ^{2}}{n}}}} \]

#### Multiclass case

Take your \(K \times K\) confusion matrix \(C\), then

\[ {\displaystyle {\text{MCC}}={\frac {\sum _{k}\sum _{l}\sum _{m}C_{kk}C_{lm}-C_{kl}C_{mk}}{{\sqrt {\sum _{k}(\sum _{l}C_{kl})(\sum _{k'|k'\neq k}\sum _{l'}C_{k'l'})}}{\sqrt {\sum _{k}(\sum _{l}C_{lk})(\sum _{k'|k'\neq k}\sum _{l'}C_{l'k'})}}}}} \]

### ROC/AUC

Receiver Operator Characteristic/Area Under Curve. Supposedly dates back to radar operators in the mid-century. HaMc83 talk about the AUC for radiology; Supposedly Spac89 introduced it to machine learning, but I havenâ€™t read the article in question. Allows you to trade off importance of false positive/false negatives.

### Cross entropy

Iâ€™d better write down form for this, since most ML toolkits are curiously shy about it.

Let \(x\) be the estimated probability and \(z\) be the supervised class label. Then the binary cross entropy loss is

\[ \ell(x,z) = -z\log(x) - (1-z)\log(1-x) \]

If \(y=\operatorname{logit}(x)\) is not a probability but a logit, then the numerically stable version is

\[ \ell(y,z) = \max\{y,0\} - y + \log(1+\exp(-|x|)) \]

### f-measure et al

đźš§

# Refs

Flach, Peter, JosĂ© HernĂˇndez-Orallo, and Cesar Ferri. 2011. â€śA Coherent Interpretation of AUC as a Measure of Aggregated Classification Performance.â€ť In *Proceedings of the 28th International Conference on Machine Learning (ICML-11)*, 657â€“64. http://www.icml-2011.org/papers/385_icmlpaper.pdf.

Gorodkin, J. 2004. â€śComparing Two K-Category Assignments by a K-Category Correlation Coefficient.â€ť *Computational Biology and Chemistry* 28 (5-6): 367â€“74. https://doi.org/10.1016/j.compbiolchem.2004.09.006.

Hand, David J. 2009. â€śMeasuring Classifier Performance: A Coherent Alternative to the Area Under the ROC Curve.â€ť *Machine Learning* 77 (1): 103â€“23. https://doi.org/10.1007/s10994-009-5119-5.

Hanley, J A, and B J McNeil. 1983. â€śA Method of Comparing the Areas Under Receiver Operating Characteristic Curves Derived from the Same Cases.â€ť *Radiology* 148 (3): 839â€“43. https://doi.org/10.1148/radiology.148.3.6878708.

Jung, Alexander, Alfred O. Hero III, Alexandru Mara, and Saeed Jahromi. 2016. â€śSemi-Supervised Learning via Sparse Label Propagation,â€ť December. http://arxiv.org/abs/1612.01414.

Lobo, Jorge M., Alberto JimĂ©nez-Valverde, and Raimundo Real. 2008. â€śAUC: A Misleading Measure of the Performance of Predictive Distribution Models.â€ť *Global Ecology and Biogeography* 17 (2): 145â€“51. https://doi.org/10.1111/j.1466-8238.2007.00358.x.

Matthews, B. W. 1975. â€śComparison of the Predicted and Observed Secondary Structure of T4 Phage Lysozyme.â€ť *Biochimica et Biophysica Acta (BBA) - Protein Structure* 405 (2): 442â€“51. https://doi.org/10.1016/0005-2795(75)90109-9.

Powers, David Martin. 2007. â€śEvaluation: From Precision, Recall and F-Measure to ROC, Informedness, Markedness and Correlation.â€ť http://dspace.flinders.edu.au/xmlui/handle/2328/27165.

Reid, Mark D., and Robert C. Williamson. 2011. â€śInformation, Divergence and Risk for Binary Experiments.â€ť *Journal of Machine Learning Research* 12 (Mar): 731â€“817. http://www.jmlr.org/papers/v12/reid11a.html.

Spackman, Kent A. 1989. â€śSignal Detection Theory: Valuable Tools for Evaluating Inductive Learning.â€ť In *Proceedings of the Sixth International Workshop on Machine Learning*, 160â€“63. San Francisco, CA, USA: Morgan Kaufmann Publishers Inc. http://dl.acm.org/citation.cfm?id=102118.102172.