Tracking physical activities performed in people's daily life are essential to quality of life. In this study, physical activities of people are tacked using wearable technology instead of traditional systems. In our study, activity classification for skiing, snowboarding, running, walking and driving activities is performed. A wristband which contains an accelerometer and a WiFi-supported Photon development card is used to collect the activity data. Collected raw data are transformed into a data set with twenty one features (mean, standard deviation, correlation, etc.) and activity classification was performed through four different classification methods (k-nearest neighbor, Naive Bayes, Random Forest and Support Vector Machines). The accuracy rates obtained for the classification are found to be in range of 90-99%. To the best of our knowledge, skiing and snowboarding activities are not used in activity recognition studies before. Our study is different from other studies in the literature by this aspect and by the use of a wristbandfor data collection stage. In the future, it is planned to design and develop a Photon-based wristband and to conduct detailed research for skiing and snowboarding activities.
