EXPERIMENTING STRESS LEVEL PREDICTION USING EVOLVING BAGGING METHOD WITH DATA STREAM
Nguyen Anh Duy, Huynh Vo Huu Tri, Duong Duy Khanh, Ngo Ho Anh Khoi
Abstract: The increasing prevalence of depression resulting from excessive stress is leading to a rising number of suicide cases among many young people today. Therefore, this is an urgent issue that requires research and implementation of appropriate interventions. This article discusses using a database collected from depression symptoms called Stress-Lysis and applying the Bagging algorithm combined with the evolving machine learning method for data stream processing to analyze and predict stress levels. The goal is to provide timely treatment methods for those affected by the condition. Experimental results demonstrate that this combination achieves favorable outcomes and has been implemented in the development of a prototype electronic platform.
Keywords: Stress-Lysis, stress level prediction, bagging, evolving machine learning, data stream.
Dataset Attributes
| Features | Description |
|---|---|
| Humidity | Humidity is typically known as the concentration of water vapor in the air, but in the specific dataset context, it refers to the humidity of the human body, also known as sweat, measured by a smart sensor device. As the body's sweat increases, the current between two electrodes also increases, turning the human body into a variable resistor. Sensors detecting humidity can be used to monitor sweat secretion, which is controlled by the human central nervous system. Monitoring the amount of sweat produced can aid in identifying stress levels and stimulation in the subject being monitored. Sweat activity functions as a variable used in many physiological feedback applications, such as lie detection and emotion recognition. Normal sweat activity is referred to as sweating, while excessive sweating disorder is called hyperhidrosis, related to emotional, occupational, and social stress. |
| Temperature | Human body temperature is expressed as a percentage. Body temperature is a major symptom for any health issues. Temperature rate is the rate of change of body temperature over a certain time period. By analyzing patterns in temperature changes, one can analyze a person's physical and mental state. The rate of body temperature change is the variation rate of body temperature over a certain time period. In general, temperature sensors can be categorized into two types: contact temperature sensors measure temperature when placed on the body, and non-contact sensors measure infrared or optical radiation received from any area of the body. In this work, we modeled a contact temperature sensor capable of tracking the rate of change of body temperature. |
| Step Count | Step count also has a significant impact, particularly on individuals experiencing severe stress, where higher step counts are observed. An accelerometer sensor device measures the rate of change in velocity of an object. They typically consist of three separate accelerometers orthogonally mounted on a physical 3-axis system (x, y, and z). Forces that cause acceleration can be static or dynamic. Sensing voltage is generated when tiny crystal structures are affected by these forces. In collecting this data, an accelerometer sensor device was used to measure an individual's step count. |
| Prediction Label (Class) | Decisive value indicating stress levels, with three possible classes: 0, 1, or 2 |
CLICK TO KNOW MORE
