What is the best way to calculate the % improvement in accuracy? For instance, if proposed technique achieve precision of 0.85 and SVM 0.78. what could be the formula to compute % improvement in accuracy achieved by proposed solution?
Dear Muhammad Ali
First: work out the difference (increase) between the two numbers you are comparing. Then: divide the increase by the original number and multiply the answer by 100. % increase = Increase ÷ Original Number × 100.
Kind Regards
Qamar Ul Islam
you can use a confusion matrix report to calculate some parameters like precision, recall, and f1 score after that use classification accuracy to calculate accuracy ( in Python)
Work on confusion matrix, folds (5, 10, 15) of cross validation and check the accuracy, you can then use recall and precision to calculate the F-measure.
Hi,
You can follow this link. Good luck
https://www.dataschool.io/simple-guide-to-confusion-matrix-terminology/
Dear Muhammad Ali
The following papers can help you to understand this
Article Adaptive Dynamic Meta-Heuristics for Feature Selection and C...
Article Novel Feature Selection and Voting Classifier Algorithms for...
Muhammad Ali
%improvement=100 - (benchmark_algorithm/your_algorithm)x100
Examples of such calculations are available in the result section of this paper.
Article Balancer Genetic Algorithm-A Novel Task Scheduling Optimizat...
- Badges
- Science topic