Shulou(Shulou.com)06/01 Report--

In this issue, the editor will bring you about how to use 4 samples to draw the ROC curve. The article is rich in content and analyzes and narrates it from a professional point of view. I hope you can get something after reading this article.

ROC curve, as an important curve to evaluate the sensitivity of machine learning model, is widely used in classification task evaluation mechanism.

But many friends still have a vague understanding of the ROC curve, thinking that the x-axis is FPR, the y-axis is TPR, and the group curve is somewhat mysterious.

Today, we use four samples and use logical regression to calculate the FRP and TPR values of the corresponding classification thresholds from 0.2 to 0.8 respectively from 0.2 to 0.8 respectively.

In this way, you will get 6 (FRP,TPR) points and form a curve, which is not a ROC curve.

# there is a supervisory task, and the actual values of the four samples:

Target = np.array ([1pc0pl 1p0pl])

Calculation formula:

Def TPR (TP,FN):

Return TP / (TP + FN)

Def FPR (FP,TN):

Return FP / (FP + TN)

By using logical regression Logistic regression, the probability distributions of four samples are obtained as follows: [0.8, 0.6, 0.4, 0.2]

When the classification threshold parameter alpha is 0.0:

Y0 = np.array ([1pm 1pm 1pm 1])

P00 = [TPR (2jin0), FPR (2jin0)]

P00

Results:

[1.0, 1.0]

When the classification threshold parameter alpha is 0.2:

# similarly, if the classification threshold alpha = 0.2,

#

Y02 = np.array ([1pm 1pm 1pm 1])

P02 = [TPR (2jin0), FPR (2jin0)]

P02

Results:

[1.0, 1.0]

When the classification threshold parameter alpha is 0.4:

Target = np.array ([1pc0pl 1p0pl])

Y04 = np.array ([1pm 1pm 1pm 0])

P04 = [TPR (2jin0), FPR (1pjin1)]

P04

Results:

[1.0, 0.5]

When the classification threshold parameter alpha is 0.6:

# similarly, if the classification threshold alpha = 0.6

#

Target = np.array ([1pc0pl 1p0pl])

Y06 = np.array ([1, 1, 1, 1, 0, 0])

P06 = [TPR (1), FPR (1)]

P06

Results:

[0.5, 0.5]

When the classification threshold parameter alpha is 0.8:

# similarly, if the classification threshold alpha = 0.8

#

Target = np.array ([1pc0pl 1p0pl])

Y08 = np.array ([1pc0p0p0p0p0pl])

P08 = [TPR (1), FPR (0)]

P08

Results:

[0.5, 0.0]

When the classification threshold parameter alpha is 1.0:

# similarly, if the classification threshold alpha = 1.0,

#

Target = np.array ([1pc0pl 1p0pl])

Y10 = np.array ([01990p0p0p0p0p0])

P10 = [TPR (0Jing 2), FPR (0Jing 2)]

P10

Results:

[0.0, 0.0]

Draw the above 6 points to get the ROC curve:

Npxy = np.array (xy)

Fpr = npxy [:, 1]

Tpr = npxy [:, 0]

Plt.scatter (fpr,tpr)

Plt.plot (fpr,tpr)

Plt.grid ()

Alphas = [0.010, 0.2, 0.4, 0.6, 0.8, 1.0]

For alpha,x,y in zip (alphas,fpr,tpr):

Plt.annotate ("alpha=%s"% (alpha,), xy= (xQuery), xytext= (- 20,10), textcoords='offset points')

The above is the editor for you to share how to use 4 samples to draw the ROC curve, if you happen to have similar doubts, you might as well refer to the above analysis to understand. If you want to know more about it, you are welcome to follow the industry information channel.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.