TrulyFitAI

trulyfit_project_ai6

Data Enrichment and Fitness recommendation Model

Timeline

Task	Duration	Deadline
EDA	26th - 1st October	2nd October
Data Enrichment	4th - 9th October	10th October
Preprocessing Pipeline	10th October - 11th October	12th October
Model Training	13th - 18th October	20th October
Front-end & Model Deployment	21st - 25th October	25th October
Documentation	Throughout the Project	End of project

Fitness Recommendation AI — End-to-End Project Documentation

1. Problem Statement

Modern fitness journeys often fail due to one-size-fits-all meal and workout plans that ignore individual differences in body type, metabolism, goals, and adherence levels.
People struggle to:

• Know how many calories they need per day to reach their goals.
• Get personalized meal and exercise plans that fit their lifestyle.
• Receive adaptive recommendations based on their real progress over time.

This project aims to solve these problems using Machine Learning (ML) and data-driven personalization.

---

2. Project Objective

To build an AI-powered fitness recommendation system that:

1. Predicts a user’s daily calorie requirement based on personal attributes and goals.
2. Generates a personalized meal plan and workout routine that align with the calorie target and goal.
3. Tracks user progress over time (weight, adherence, energy level) to adapt recommendations dynamically.

The end goal is to provide an intelligent, continuously learning fitness assistant.

---

3. Methodology Overview

The system will be developed in phases, evolving from a rule-based MVP to an adaptive ML system.

Phase	Focus	Description
Phase 1	Data Setup	Gather and structure user profile, meal, and workout datasets.
Phase 2	Calorie Prediction	Build a regression model to predict daily calorie needs based on user profile.
Phase 3	Personalized Recommendation	Generate meals and workouts matching calorie goals and user preferences.
Phase 4	Progress Tracking	Collect feedback such as adherence, weight change, and energy level.
Phase 5	Adaptive Learning	Retrain model weekly using progress data to make adaptive calorie/workout recommendations.
Phase 6	Multi-User Intelligence	Use collaborative filtering to recommend meals/workouts based on similar users’ success.
Phase 7	Deployment	Build a Streamlit dashboard for real-time interaction and visualization.

---

4. Data Requirements and Sources

A. User Profile Dataset

Purpose: To predict calorie needs and personalize recommendations.

Feature	Type	Description
user_id	ID	Unique identifier
name	string	User name
age	numeric	User’s age
gender	categorical	Male/Female
height_cm	numeric	Height in cm
weight_kg	numeric	Weight in kg
goal	categorical	lose_weight / maintain / gain_muscle
bmi	numeric	Calculated as weight / (height²)
experience_level	categorical	beginner / intermediate / advanced
equipment	categorical	home / gym / none
calorie_target	numeric	(Label) Daily calorie target — predicted or derived

Source: Provided 20k user profile dataset (Final_data.csv)

---

B. Meal Dataset

Purpose: To generate balanced, goal-aligned meal recommendations.

Feature	Description
meal_id	Unique meal identifier
meal_name	Meal name (e.g., "Grilled Chicken with Rice")
calories	Total calories per portion
protein	Protein content (g)
carbs	Carbohydrates (g)
fats	Fat content (g)
category	e.g., breakfast/lunch/dinner/snack
goal_tag	lose_weight / gain_muscle / maintain
source	e.g., USDA API / manually curated

Source: USDA FoodData Central API (for open-source nutrition data)

---

C. Exercise Dataset

Purpose: To recommend workouts aligned with goal, equipment, and level.

Feature	Description
exercise_id	Unique ID
exercise_name	e.g., "Push-ups"
target_muscle	e.g., chest, legs
difficulty	beginner / intermediate / advanced
equipment	bodyweight / dumbbell / barbell / none
duration_min	Average duration
calories_burned	Estimated calories burned per session

Source: ExerciseDB API or curated GYM.csv dataset

---

5. System Architecture

    A[User Profile Input] --> B[Calorie Prediction Model]
    B --> C[Calorie Target (kcal/day)]
    C --> D[Meal Plan Generator]
    C --> E[Workout Plan Generator]
    D & E --> F[Personalized Recommendation Output]
    F --> G[User Progress Logging]
    G --> H[Adaptive Learning Model]
    H --> D & E