Monitoring Student Activities with Smartwatches: On the Academic Performance Enhancement.

Area of Science:

• Human-Computer Interaction
• Machine Learning
• Educational Technology

Background:

• Understanding student habits is crucial for academic performance.
• Wearable technology offers new avenues for tracking daily student activities.
• Previous research has not fully explored the link between sensor data and academic outcomes.

Purpose of the Study:

• To investigate the relationship between student daily activities and academic performance.
• To develop a system for recognizing student activities using smartwatch data.
• To lay the groundwork for a recommendation system to enhance student academic success.

Main Methods:

• Utilized smartwatches and an Android application to collect data on undergraduate students' daily activities.
• Employed supervised machine learning algorithms for activity recognition.
• Applied discrete wavelet transform for feature extraction from gyroscope and accelerometer signals to enhance classification accuracy.

Main Results:

• Achieved satisfactory activity recognition accuracy of 86.9% using the Random Forest algorithm.
• Demonstrated a strong correlation between smartwatch sensor signals and recognized daily living activities.
• Validated the feasibility of using sensor data for understanding student behavior patterns.

Conclusions:

• Smartwatch sensor data can reliably recognize student activities.
• This approach supports the potential for developing automatic activity-labeling and pattern recognition systems.
• Future work can lead to recommendation systems aimed at improving student academic performance.