Skip to content

Analysis

Jump to bottom
Faheem Ershad edited this page May 14, 2018 · 5 revisions

The guide below shows how to compare the training session and online prediction data from the trained linear SVM classifiers. This should be done immediately after completing the training and online prediction sessions with the cursor control. Again, this is for the Two Binary Classifiers version, but the procedure for Left/Right version is the same.

  1. Download all the files in the Two Binary Classifiers folder in the Online-Training-and-Prediction branch, if you haven't already done so. Open the Online_prediction_analysis_NLR.m script. Run the script.

  2. There will be four plots that show up in this case. Figures 1 and 2 will be plots for the Not moving/moving classifier and Left/Right classifier (respectively) during the training session. These are shown below. If you scroll down to the end of the script, you will see variables f1 and f2 which can be modified to compare any two features of your choice. Note that the colors here are from predicted and NOT true labels.

  3. Figures 3 and 4 are plots for Not moving/moving classifier and Left/Right classifier (again, respectively) during the online prediction. You will need to adjust the legend since it doesn't have all the points. Do this by disabling the legend and re-enabling it: the button is indicated in the screenshot below. Again, you can modify the f1 and f2 variables to values of your choice.

Clone this wiki locally