I am testing my algorithm on 20 different datasets. So, as a result I derived 20 F1 scores based on precision and recall for each run (dataset). Is there any way to combine all these 20 F1 scores into one metrics or summary statistics to report how the algorithm performed on the 20 datasets?
Weighted Mean: If some datasets are more important than others, you can assign weights to each dataset and calculate a weighted average.
Statistical Tests: You can perform statistical tests to determine if there is a statistically significant difference in F1 scores across the datasets. This can help identify if your algorithm consistently performs better on certain types of datasets.
Box Plots or Violin Plots: Use visualizations to show the distribution of F1 scores across datasets. This can help identify any outliers or trends.
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts