Figure 2
From: Classification of human walking context using a single-point accelerometer
Classification model—We extracted walking periods from the accelerometer that we then labeled using self-report and GPS data. We extracted features from the accelerometer data and the stride detection. Labeled walking periods were then split into training, validation, and testing set. We trained and validated two different learning algorithms, Random Forest and Ensemble SVM, using a leave-one-participant-out scheme. This led to 15 trained models that were then tested on the 5 remaining untouched participants’ data. The best-performing model was chosen for the rest of the analyses.
