PREDICTION SYSTEM DEVELOPMENT OF AN IOT-BASED MOUNTAIN HIKER MONITORING DEVICE

• Main Author: Ichsandro D Noor, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/73863
• Institution: Institut Teknologi Bandung

Mountain climbing is an activity that is identic with a long and risky journey with the main goal of reaching the top of the mountain. There are several main problems in this activity, namely health risks related to Acute Mountain Sickness (AMS) and the risk of getting lost from the predetermined path while climbing. Therefore, to improve the experience and safety of mountaineering activities, we are developing solutions in the form of Internet of Things (IoT) and Machine Learning-based devices that can track the mountain climber’s location by the system manager and monitor their health conditions in real time and can give an alert when it is detected to have a health risk or lost. Overall system development was carried out using the V-Model development method and specifically for Machine Learning development it was carried out using the CRISP-DM method with the main result being a "Proof of Concept" in the form of a hardware system used by mountain climbers, an interface system on web applications, and a prediction and classification system using Machine Learning models. As the result, the classification system was built using Random Forest model which is able to classify the climber's vital condition into 4 levels of health risks related to Acute Mountain Sickness and the prediction system built using SVR Polynomial which is able to predict the climber's vital condition in the next climbing post. After being tested with the experts, it was concluded that the system can detect and minimize health risks with a prevention that can be carried out by system managers based on prediction results and give an alert to climbers based on the classification results.