BEEG performance measurement is calculated using the general method as well as the “standard method” of Confusion Matrix and is presented in following sections.
6.2.1 Using General Method: Although more or less, the end results were similar for all mental tasks performed on subjects under various categories as shown in all graphs of Chapter 6 and Fig B.1 to B.10 in appendix B, but for ease, mean accuracy chart is figured out here using all graphs and are shown next. The mean accuracy of BEEG using Confusion Matrix method (can be considered as standard method as most BCIs have used it for performance measure) is separately calculated in Chapter 6.2.2. The table below shows the overall performance % of all samples per subject.
Table 6.60: …show more content…
This method is also incorporated here for BEEG performance measure as till date no standard method is available for measuring BCI accuracy but most of the BCIs have used this method. Therefore BEEG is also measured through confusion matrix for four class classifier and performance constructs are compared with five other BCIs in chapter 7. For better understanding, the Confusion Matrix of two class classifier is shown in fig 6.55. Predicted -ve +ve
Actual -ve a b +ve c d
Fig 6.56: Confusion Matrix for two class Classifier [4.25]
The description of Confusion Matrix for two class classifier is given as below [4.25]:
• a is the number of correct predictions that an instance is negative
• b is the number of incorrect predictions that an instance is …show more content…
All these readings are then put into the confusion matrix as described above. The confusion matrix with all such readings put for four subjects “1A”, 2B”, “3C” and “4D” is given in following figures.
Confusion matrix for BEEG for above mentioned subjects and samples is shown in following figures:
Confusion Matrix % Apple Run Music File Open Subject’s Accuracy (%)
Apple 25 2 0 1 92.857
Run 5 20 3 0 71.429
Music 2 1 22 3 78.571
File Open 1 0 0 27 96.429
Fig 6.57: Confusion Matrix Result analysis for subject “1A”
Overall Accuracy for Subject “1A” calculated using Eq. 6.2.1 is: 84.821%
Confusion Matrix % Apple Run Music File Open Subject’s Accuracy (%)
Apple 27 0 0 1 96.429
Run 2 23 2 1 82.143
Music 1 1 24 2 85.714
File Open 1 1 0 26 92.857
Fig 6.58: Confusion Matrix Result analysis for subject “2B”
Overall Accuracy for Subject “2B” using Eq. 6.2.1 is: 89.286%
Confusion Matrix % Apple Run Music File Open Subject’s Accuracy (%)
Apple 27 0 1 0 96.429
Run 3 24 1 0 85.714
Music 4 2 19 3 67.857
File Open 2 1 1 24 85.714
Fig 6.59: Confusion Matrix Result analysis for subject “3C”
Overall Accuracy for Subject “3C” using Eq. 6.2.1 is:
6.2.1 Using General Method: Although more or less, the end results were similar for all mental tasks performed on subjects under various categories as shown in all graphs of Chapter 6 and Fig B.1 to B.10 in appendix B, but for ease, mean accuracy chart is figured out here using all graphs and are shown next. The mean accuracy of BEEG using Confusion Matrix method (can be considered as standard method as most BCIs have used it for performance measure) is separately calculated in Chapter 6.2.2. The table below shows the overall performance % of all samples per subject.
Table 6.60: …show more content…
This method is also incorporated here for BEEG performance measure as till date no standard method is available for measuring BCI accuracy but most of the BCIs have used this method. Therefore BEEG is also measured through confusion matrix for four class classifier and performance constructs are compared with five other BCIs in chapter 7. For better understanding, the Confusion Matrix of two class classifier is shown in fig 6.55. Predicted -ve +ve
Actual -ve a b +ve c d
Fig 6.56: Confusion Matrix for two class Classifier [4.25]
The description of Confusion Matrix for two class classifier is given as below [4.25]:
• a is the number of correct predictions that an instance is negative
• b is the number of incorrect predictions that an instance is …show more content…
All these readings are then put into the confusion matrix as described above. The confusion matrix with all such readings put for four subjects “1A”, 2B”, “3C” and “4D” is given in following figures.
Confusion matrix for BEEG for above mentioned subjects and samples is shown in following figures:
Confusion Matrix % Apple Run Music File Open Subject’s Accuracy (%)
Apple 25 2 0 1 92.857
Run 5 20 3 0 71.429
Music 2 1 22 3 78.571
File Open 1 0 0 27 96.429
Fig 6.57: Confusion Matrix Result analysis for subject “1A”
Overall Accuracy for Subject “1A” calculated using Eq. 6.2.1 is: 84.821%
Confusion Matrix % Apple Run Music File Open Subject’s Accuracy (%)
Apple 27 0 0 1 96.429
Run 2 23 2 1 82.143
Music 1 1 24 2 85.714
File Open 1 1 0 26 92.857
Fig 6.58: Confusion Matrix Result analysis for subject “2B”
Overall Accuracy for Subject “2B” using Eq. 6.2.1 is: 89.286%
Confusion Matrix % Apple Run Music File Open Subject’s Accuracy (%)
Apple 27 0 1 0 96.429
Run 3 24 1 0 85.714
Music 4 2 19 3 67.857
File Open 2 1 1 24 85.714
Fig 6.59: Confusion Matrix Result analysis for subject “3C”
Overall Accuracy for Subject “3C” using Eq. 6.2.1 is: