HabitCheckAI: A Machine Learning Based Web Dashboard for Smartphone Habit and Health Analysis

Abstract

HabitCheckAI is a Flask based web application for reviewing everyday habit and health related data in one place. It reads user records from Firebase Realtime Database, accepts values through browser forms, and presents results with JavaScript based pages and charts. The project contains four serialized scikit learn models for sleep score, study and productivity score, stress category, and overall health score. Four CSV files with 2,000 records each support the training code. Sleep, study, and health use Random Forest regression; stress uses Random Forest classification. The dashboard also provides weekly summaries, simple recommendations, wellness indicators, and rule based behavior insights. This paper describes HabitCheckAI as a software prototype. It shows how a Flask application, Firebase data retrieval, model artifacts, and a browser interface can be connected for habit tracking. The available project material does not include a confirmed source for the training data, an external validation study, or a stored record of model results. For that reason, no accuracy or error value is presented as a verified result. The paper also records three issues that affect the current build: the deployed study score does not use the trained Random Forest output, stress inputs use inconsistent units in different components, and the login pages do not match the active backend login flow. Keywords: habit monitoring, smartphone behavior, Flask, Firebase Realtime Database, Random Forest, sleep analysis, stress classification, health dashboard