i want to ask how can i convert True Positive number into correctly classified instances. for example i have test data total number (31029) and its Dos class have total number(229853). When we apply any Tree Algorithm in WEKA it give TP = 0.998 and FP = 0.002. So how can we convert it in number or classified instances. i am attaching any image for relevant paper also
Though, i have difficulty in understanding your question but if i can get you, you wish to know the number of instances that gives that your true positive TP of 0,998 and False positive FP of 0.002. If this is your question, you need to know the total test data from the KDD data set. The instance you are looking for divided by the total will give you the TP or FP as the case may be. So, number of instance therefore will be the total multiply by the TP or FP as the case may be. And should you be considering say a binary class situation, your accuracy computation is not yet completed until you have also talked of the true negative TN and False negative FN. You may then get your overall accuracy from the formula (TP+TN)/(TP+TN+FP+FN). if you want it in percentage, just multiply by 100. I hope this helps, but if not try and be more specific or try to ask again in a more clearer form. Wishing you all the best.
Normally, your TP rate or confusion matrix shows the performance of your model on the test data. In that sense I am a bit confused how your test data has less instances than your Dos class.
I am trying to give the answer specifically for your figure from weka.
If you see your figure, for class "a" (i.e., the Dos class), the TP rate is computed by: 229407/(229407+196+61+0+189) = 0.998....
In the confusion matrix in your figure, you already have the information regarding the correctly classified instances.
For example, there are 229407 instances of "a" are classified correctly.
The diagonal entries in your confusion matrix (in the figure) contain the correctly classified instances for the corresponding class shown in the column name. It means that, for b or Probe, it is 3792, for c (i.e., r2l) it is 12923, and so on. Moreover, if you want to know the total test instances for a particular class, just add the numbers in the corresponding rows, e.g., first row for a, second row for b, and so on.
On the other hand, if you want to directly calculate the no. of correctly classified instances for the positive class, then you have to know the total test instances of the positive class and and multiply it with the TP rate.
For more clarification about the confusion matrix terminologies, you read this web page: http://www.dataschool.io/simple-guide-to-confusion-matrix-terminology/
Dear,
You can refer this article:
Tien Bui, D.; Pradhan, B.; Lofman, O.; Revhaug, I. Landslide Susceptibility Assessment in Vietnam Using Support Vector Machines, Decision Tree, and Naive Bayes Models. Mathematical Problems in Engineering. 2012:1-26; 2012
You can get the True positive from the confusion matrix. WEKA gives confusion matrix. Diagonal elements are TP values.
- Badges
- Science topic
- Similar topics
- Computer Science
- Data Mining