24
$\begingroup$

In the discussion : how to generate a roc curve for binary classification, I think that the confusion was that a "binary classifier" (which is any classifier that separates 2 classes) was for Yang what is called a "discrete classifier" (which produces discrete outputs 0/1 like an SVM) and not continuous outputs like ANN or Bayes classifiers ... etc. So, the discussion was about how the ROC is plotted for "binary continuous classifiers", and the answer is that the outputs are sorted by their scores since the outputs are continuous, and a threshold is used to produce each point on the ROC curve.

My question is for "binary discrete classifiers", such as SVM, the output values are 0 or 1. So the ROC produces just one point and not a curve. I'm confused as to why we still call it a curve?!! Can we still talk about thresholds? How can one use thresholds in SVM in particular? How can one compute the AUC?, Does cross-validation play any role here?

$\endgroup$
  • 9
    $\begingroup$ An SVM outputs a real decision value, namely the signed distance to the separating hyperplane in feature space. In classification the label is assigned based on the sign of this decision value. As such, SVMs do output more than just a binary value, their output is just binarized as a post-processing step in classification. $\endgroup$ – Marc Claesen Feb 17 '14 at 11:11
14
$\begingroup$
  • Yes, there are situations where the usual receiver operating curve cannot be obtained and only one point exists.

  • SVMs can be set up so that they output class membership probabilities. These would be the usual value for which a threshold would be varied to produce a receiver operating curve.
    Is that what you are looking for?

  • Steps in the ROC usually happen with small numbers of test cases rather than having anything to do with discrete variation in the covariate (particularly, you end up with the same points if you choose your discrete thresholds so that for each new point only one sample changes its assignment).

  • Continuously varying other (hyper)parameters of the model of course produces sets of specificity/sensitivity pairs that give other curves in the FPR;TPR coordinate system.
    The interpretation of a curve of course depends on what variation did generate the curve.

Here's a usual ROC (i.e. requesting probabilities as output) for the "versicolor" class of the iris data set:

  • FPR;TPR (γ = 1, C = 1, probability threshold):
    ROC

The same type of coordinate system, but TPR and FPR as function of the tuning parameters γ and C:

  • FPR;TPR (γ, C = 1, probability threshold = 0.5):
    gamma

  • FPR;TPR (γ = 1, C, probability threshold = 0.5):
    cost

These plots do have a meaning, but the meaning is decidedly different from that of the usual ROC!

Here's the R code I used:

svmperf <- function (cost = 1, gamma = 1) {
    model <- svm (Species ~ ., data = iris, probability=TRUE, 
                  cost = cost, gamma = gamma)
    pred <- predict (model, iris, probability=TRUE, decision.values=TRUE)
    prob.versicolor <- attr (pred, "probabilities")[, "versicolor"]

    roc.pred <- prediction (prob.versicolor, iris$Species == "versicolor")
    perf <- performance (roc.pred, "tpr", "fpr")

    data.frame (fpr = perf@x.values [[1]], tpr = perf@y.values [[1]], 
                threshold = perf@alpha.values [[1]], 
                cost = cost, gamma = gamma)
}

df <- data.frame ()
for (cost in -10:10)
  df <- rbind (df, svmperf (cost = 2^cost))
head (df)
plot (df$fpr, df$tpr)

cost.df <- split (df, df$cost)

cost.df <- sapply (cost.df, function (x) {
    i <- approx (x$threshold, seq (nrow (x)), 0.5, method="constant")$y 
    x [i,]
})

cost.df <- as.data.frame (t (cost.df))
plot (cost.df$fpr, cost.df$tpr, type = "l", xlim = 0:1, ylim = 0:1)
points (cost.df$fpr, cost.df$tpr, pch = 20, 
        col = rev(rainbow(nrow (cost.df),start=0, end=4/6)))

df <- data.frame ()
for (gamma in -10:10)
  df <- rbind (df, svmperf (gamma = 2^gamma))
head (df)
plot (df$fpr, df$tpr)

gamma.df <- split (df, df$gamma)

gamma.df <- sapply (gamma.df, function (x) {
     i <- approx (x$threshold, seq (nrow (x)), 0.5, method="constant")$y
     x [i,]
})

gamma.df <- as.data.frame (t (gamma.df))
plot (gamma.df$fpr, gamma.df$tpr, type = "l", xlim = 0:1, ylim = 0:1, lty = 2)
points (gamma.df$fpr, gamma.df$tpr, pch = 20, 
        col = rev(rainbow(nrow (gamma.df),start=0, end=4/6)))

roc.df <- subset (df, cost == 1 & gamma == 1)
plot (roc.df$fpr, roc.df$tpr, type = "l", xlim = 0:1, ylim = 0:1)
points (roc.df$fpr, roc.df$tpr, pch = 20, 
        col = rev(rainbow(nrow (roc.df),start=0, end=4/6)))
$\endgroup$
  • $\begingroup$ This is very clear, Thanks. I had this idea that varying b is like moving the hyperplane and thus having different (TPR, FPR)! but with the SVM classifier, the bias b is learned, so it seems to be the best parameter? no?, if this is the case, there is no need to perform the ROC analysis?, no? $\endgroup$ – Abdelhak Mahmoudi Sep 24 '12 at 10:43
  • $\begingroup$ @AbdelhakMahmoudi: I guess parameters learned by the model are not what you want to vary. But as you can have a probability output (I did not dig into the code, so I don't know whether the probability SVM are really equvalent to the "hard" ones), why not use it? This is a very common outcome from which the ROC is generated. R's svm function is an interface to the widely used libsvm, so you are not restricted to use R to do this. $\endgroup$ – cbeleites Sep 24 '12 at 19:48
8
$\begingroup$

Normally, the predicted label $\hat{y}$ from SVM is given by $\hat{y}=\mbox{sign}({\mathbf w^T x}+b)$, where ${\mathbf w}$ is the SVM-optimized weights of the hyper-plane, and the $b$ is the SVM-optimized intercept. This can also be re-written as follows: \begin{eqnarray} \hat{y} & = & \left\{\begin{array}{cc} 0 & \mbox{if}~~{\mathbf w^T x}+b < 0 \\ 1 & \mbox{otherwise} \end{array} \right. \end{eqnarray}

However, if we introduce a threshold $\eta$, we can control the positive detection rate by varying $\eta$, i.e. \begin{eqnarray} \hat{y} & = & \left\{\begin{array}{cc} 0 & \mbox{if}~~{\mathbf w^T x}+b < \eta \\ 1 & \mbox{otherwise} \end{array} \right. \end{eqnarray}

By varying $\eta$, we can produce an ROC using SVM, and thereby adjusting the sensitivity and specificity rate.

For example, if we want to do it in python, we can extract ${\mathbf w}$ and $b$ using threshold $\eta$ as follows.

>>> from sklearn.svm import SVC
>>> model = SVC(kernel='linear', C=0.001, probability=False, class_weight='balanced')
>>> model.fit(X, y)
>>> # The coefficients w are given by
>>> w = list(model.coef_)
>>> # The intercept b is given by
>>> b = model.intercept_[0]
>>> y_hat = X.apply(lambda s: np.sum(np.array(s)*np.array(w))+b, axis=1)
>>> y_hat = (y_hat > eta).astype(float)
$\endgroup$
3
$\begingroup$

The ROC curve plots specificity vs sensitivity which varies with threshold of a covariate (which may be continuous or discrete). I think you are confusing the covariate with the response and perhaps do not fully understand what an ROC curve is. It is certainly a curve if the covariate is continuous and we look at a threshold for the covariate changing continuously. If the covariate is discrete you can still plot is as a function of a continuous threshold. Then the curve would be flat with steps up (or down) at thresholds that correspond to the discrete values of the covariate. So this would apply to SVM and any other discrete classifiers.

Regarding AUC since we still have an ROC (an estimated one) we can still compute the area under it. I am not sure whay you had in mind with your question about cross-validation. In the context of classification problems cross-validation is used to get unbiased or nearly unbiased estimates of error rates for the classifier. So it can enter into how we estimate the points on the ROC.

$\endgroup$
  • 1
    $\begingroup$ Ok, What could be the thresholds for svm classifier? $\endgroup$ – Abdelhak Mahmoudi Sep 22 '12 at 14:34
  • $\begingroup$ I don't know. What are the covariates? If you have one covariate any value could be a threshold. If you have more than one covariate the classifier performance depends on the choice of several values rather than a single threshold but it still varies in the space of the covariates. $\endgroup$ – Michael Chernick Sep 22 '12 at 15:19
  • $\begingroup$ For example, linear SVMs are based on the separating hyperplane which depends on the chosen C value (low C tolerates more training errors), could a set of C values be the the set of thresholds? $\endgroup$ – Abdelhak Mahmoudi Sep 22 '12 at 15:50
  • $\begingroup$ Yes and aren't the C values just linear combinations of the covariates? $\endgroup$ – Michael Chernick Sep 22 '12 at 15:52
  • 2
    $\begingroup$ C is the penalty factor introduced to control the trade-off between the hyperplane complexity and training errors. An other alternative could be the use of the bias b as a threshold since b is the distance from the centre of the feature space to the separating hyperplane. So varying b is like moving the hyperplane and thus having different TP and FP! This is my understanding! $\endgroup$ – Abdelhak Mahmoudi Sep 22 '12 at 17:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.