This is the data and the source code used in the paper "Emotion Recognition Using Smart Watch Sensor Data: Mixed-Design Study" published in JMIR Mental Health. If you use our code or dataset, cite the following paper:

Quiroz JC, Geangu E, Yong MH
Emotion Recognition Using Smart Watch Sensor Data: Mixed-Design Study
JMIR Ment Health 2018;5(3):e10153
URL: https://mental.jmir.org/2018/3/e10153
DOI: 10.2196/10153
PMID: 30089610

Our preliminary results were published in:

Juan C. Quiroz, Min Hooi Yong, and Elena Geangu. 2017. 
Emotion-recognition using smart watch accelerometer data: preliminary findings. 
In Proceedings of the 2017 ACM International Joint Conference on Pervasive and 
Ubiquitous Computing and Proceedings of the 2017 ACM International Symposium 
on Wearable Computers (UbiComp '17). ACM, New York, NY, USA, 805-812. 
DOI: https://doi.org/10.1145/3123024.3125614

Data was collected from 50 participants. Coding details are available in user_study_encoding.csv.

The code was written in python2.7. To install the requiements run:

pip install -r requirements.txt

Extract the accelerometer data from the recorded walking times.

python get_walking_data.py user_study_encoding.csv raw_data/ walking_data/

Extract features from sliding windows.

python extract_windows.py walking_data/m* features/

Compute classification accuracies of happy vs sad.

python user_lift.py -mo features/features_mo_ew* -mu features/features_mu_ew* -mw features/features_mw_ew* -o acc_f1

Compute classification accuracies of happy vs sad vs neutral.

python2 user_lift.py -mo features/features_mo_ew* -mu features/features_mu_ew* -mw features/features_mw_ew* -o neutral --neutral

Run permutation test to determine if accuracies are higher than baseline.

python permute_test.py -mo mo_lift_scores_log.yaml -mu mu_lift_scores_log.yaml -mw mw_lift_scores_log.yaml

Generate plot of feature importances.

python feature_importance_plot.py -mo mo_feature_import_acc_f1.yaml -mu mu_feature_import_acc_f1.yaml -mw mw_feature_import_acc_f1.yaml