Human Activity Dataset Generation - Smartphone-Based Human Activity Recognition - page 71

Table 4.8 Confusion matrices for the D 2 and D 3 dataset using the multiclass SVMwith Gaussian
kernel (MC-GK-SVM)
Activity

WK WU WD SI ST LD Sensitivity (%) Specificity (%)

MC-GK-SVM— D 2
WK 486 6 4 0 0 0 97.98 99.31

WU 12 458 1 0 0 0 97.24 98.59

WD 5 27 388 0 0 0 92.38 99.80

SI 0 2 0 450 39 0 91.65 99.71

ST 0 0 0 7 525 0 98.68 98.39

LD 0 0 0 0 0 537 100.00 100.00

Accuracy 96.50

Activity WK WU WD SI ST LD PT Sensitivity (%) Specificity (%)

MC-GK-SVM— D 3
WK 539 1 3 2 0 0 1 98.72 98.93
WU 28 513 2 1 1 0 14 91.77 99.65
WD 2 5 498 0 4 0 0 97.84 99.84
SI 0 3 0 486 68 0 0 87.25 99.24
ST 1 0 0 19 591 0 1 96.57 97.65
LD 0 0 0 0 0 604 0 100.00 100.00
PT 3 2 0 2 0 0 322 97.87 99.53
Accuracy 95.61
Note The bold diagonal highlights the most important part of the confusion matrix.

4.4.1.1 Dataset Publication

The HAR dataset has been made available for public use. It is composed of the
experiment raw inertial data and also of the processed feature vectors for eachwindow
sample. A first version (with data from D 2 ) was submitted as the Human Activity

Recognition using Smartphones dataset in the UCI Machine Learning Repository
(Bache and Lichman 2013 ). The details of the dataset are shown in the following
reference:
•

Reyes-Ortiz et al. ( 2013 ) Jorge-Luis Reyes-Ortiz, Davide Anguita, Alessandro
Ghio, Luca Oneto, and Xavier Parra. Human activity recognition using smart-
phones data set. http://archive.ics.uci.edu/ml/datasets/Human+Activity+Recogniti
on+Using+Smartphones

D 3 data is also public. It is available in
the Smartlab Laboratory website as displayed in the following reference:

Moreover, a second version based on the

• Reyes-Ortiz et al. ( 2014 ) Jorge-Luis Reyes-Ortiz, Davide Anguita, Alessandro

Ghio, Luca Oneto, and Xavier Parra. Recognition of basic activities and postural
transitions using smartphones data set. http://www.har.smartlab.ws