Skip to main content

Table 4 Confusion matrices for Binary Decision Tree, Random Forest, and Support Vector Machine classifiers trained on combined hip and wrist data

From: Machine learning algorithms for activity recognition in ambulant children and adolescents with cerebral palsy

Activity Class

Binary Decision Tree

Observed

Prediction

SED

SUM

CW

BW

 1. SED

1284 [0.94]

80 [0.06]

0 [0.00]

0 [0.00]

 2. SUM

45 [0.03]

1243 [0.91]

70 [0.05]

6 [0.00]

 3. CW

4 [0.01]

62 [0.08]

547 [0.70]

164 [0.21]

 4. BW

10 [0.01]

74 [0.05]

185 [0.12]

1285 [0.83]

Random Forest

 1. SED

1326 [0.97]

36 [0.03]

0 [0.00]

2 [0.00]

 2. SUM

20 [0.01]

1317 [0.97]

11 [0.01]

16 [0.01]

 3. CW

4 [0.01]

52 [0.07]

511 [0.66]

210 [0.27]

 4. BW

10 [0.01]

43 [0.03]

164 [0.11]

1337 [0.86]

Support Vector Machine

 1. SED

1335 [0.98]

27 [0.02]

1 [0.00]

1 [0.00]

 2. SUM

34 [0.02]

1316 [0.96]

7 [0.01]

7 [0.01]

 3. CW

4 [0.01]

71 [0.09]

529 [0.68]

173 [0.22]

 4. BW

7 [0.00]

50 [0.03]

176 [0.11]

1321 [0.85]

  1. Numbers represent observation counts. Percentage of observations for a given class reported in brackets. Values in bold face indicate number and proportion of observations within each class correctly classified
  2. SED sedentary, SUM standing utilitarian movements, CW comfortable walk, BW brisk walk