• Version
• Download 6
• File Size 391.03 KB
• File Count 1
• Create Date 14 December 2024
• Last Updated 14 December 2024

SMART WATCH DATA ANALYSIS USING PYTHON AND HUMAN HEALTH PREDICTION

1st R.Ramakrishnan 1*, 2nd P. Angarika2

1Associate Professor, Department of computer Applications, Sri Manakula Vinayagar Engineering College (Autonomous), Puducherry 605008, India

2Post Graduate student, Department of computer Applications, Sri Manakula Vinayagar Engineering                              College (Autonomous),   Puducherry 605008, India

angarikap30@gmail.com

Abstract: This project leverages smartwatch fitness data to predict health patterns and monitor daily activity trends, underscoring the role of wearables in personal health management. Using Python, Pandas, and Plotly, it handles data preprocessing, visualization, and predictive analysis on metrics such as step counts, calories burned, and active minutes. Data preprocessing includes managing missing values and standardizing the "Activity Date" field. Descriptive statistics and visualizations, including scatter plots, pie charts, and bar charts, uncover trends and behavioral patterns. Descriptive statistics provide insight into data distribution, while visualizations reveal significant trends. Scatter plots highlight correlations, such as between calories burned and steps taken, pie charts depict activity time allocation, and bar charts present active minutes across different days. These visualizations uncover behavioral patterns and emphasize data-driven insights. For predictive analysis, a Random Forest model is applied to forecast "very active minutes," representing high-intensity activity. Key predictive features include steps and calories burned, which strongly correlate with active minutes. The model achieved an accuracy of 80%, and validation metrics, such as MSE and R², confirmed its reliability. This predictive capability offers users actionable insights for fitness improvement, helping them set realistic goals and monitor progress effectively. In conclusion, this study illustrates the practical applications of machine learning in wearable data analysis, showing potential for integration into fitness-tracking apps. The model’s insights support both short-term fitness and long-term health, with future improvements including additional metrics, like heart rate and sleep data, for comprehensive health monitoring.

KEYWORDS: Smartwatch data analysis, very active minutes, Physical activity prediction, Random Forest algorithm, Personalized fitness monitoring, High-intensity activity, Machine learning in health, Predictive modeling.