Update README.md

README.md

@@ -12,81 +12,66 @@ short_description: Streamlit app with BodyFat (%) predictor
 license: mit
 ---
 
-# 🧍‍♂️ BodyFat Predictor – Ridge Regression
+# 🧍‍♂️ Body Density Predictor – Ridge Regression
 
-This project demonstrates a **machine learning regression pipeline** to predict **body density** from anthropometric measurements and convert it into **body fat percentage** using the **Siri equation**.  
-The implementation follows the original **ML Olympiad – Predicting Wellness** competition setup.
+This project presents a **machine learning regression application** that predicts **human body density** from anthropometric measurements using a **Ridge Regression** model.  
+The implementation follows the original **ML Olympiad – Predicting Wellness** competition setup, where **body density** is the primary target variable.
 
 ---
 
 ## 📌 Project Overview
 
-The goal of this project is to:
+The main objectives of this project are:
 
-- Predict **body density** using a trained **Ridge Regression** model  
-- Convert predicted density into **body fat percentage** using a physiological formula  
-- Provide a **transparent and interactive demo** via Streamlit  
-- Highlight model limitations caused by small datasets and narrow target ranges  
+- Predict **body density (Density)** from physical body measurements  
+- Build a **stable and interpretable regression model** for a small dataset  
+- Provide an **interactive Streamlit demo** for single and batch predictions  
+- Clearly communicate model assumptions and limitations  
 
-This Space is intended as an **educational and portfolio project**, not a medical tool.
+This project is intended for **educational and portfolio purposes**.
 
 ---
 
 ## 📊 Dataset & Target
 
 - **Input features:**  
-  Anthropometric measurements such as age, weight, height, and body circumferences (neck, abdomen, hip, thigh, etc.)
+  Anthropometric measurements such as age, height, weight, and body circumferences (neck, chest, abdomen, hip, thigh, etc.)
 
-- **Target variable (model output):**  
-  `Density` (body density)
+- **Target variable:**  
+  `Density` (human body density)
 
-- **Derived output (post-processing):**  
-  `BodyFat (%)` calculated using the **Siri equation**
+The model is trained **directly on Density**, without predicting body fat percentage.
 
 ---
 
-## 🧮 Body Fat Calculation (Siri Formula)
+## 🧠 Model Description
 
-After predicting body density, body fat percentage is calculated as:
-
-BodyFat (%) = 495 / Density − 450
-
-yaml
-Code kopieren
-
-This formula is widely used in body composition analysis and was applied **exactly as defined in the competition**.
+- **Model type:** Ridge Regression (L2-regularized linear regression)
+- **Reason for model choice:**
+  - Handles strong multicollinearity between body measurements
+  - Performs well on small datasets
+  - Produces stable and smooth predictions
+  - Easy to interpret and deploy
 
 ---
 
-## ⚠️ Important Note on Model Behavior
-
-- Human body density typically lies in a **very narrow range (≈ 0.95–1.10)**  
-- Small prediction errors in density can lead to **large changes in body fat percentage**
-- In some cases, valid inputs may result in **non-physiological values (e.g. negative body fat)**
+## 🧮 Feature Engineering
 
-### How this demo handles it:
-- Raw body fat values are **calculated transparently**
-- A **warning** is shown if predicted density is outside the typical human range
-- For display purposes only, body fat is **clipped to 0–60%**
-- Raw values remain visible for transparency
+Only features that were used during training are applied in the app:
 
-This behavior reflects a **known limitation** of regression models combined with physical formulas.
+- **Waist-to-Hip Ratio:**  
+Waist_hip = Abdomen / Hip
 
----
+markdown
+Code kopieren
 
-## 🧠 Model Details
+- **Body Index (BMI, imperial units):**  
+Body_Index = 703 × Weight / Height²
 
-- **Model type:** Ridge Regression (L2-regularized linear regression)
-- **Why Ridge?**
-  - Handles multicollinearity between highly correlated body measurements
-  - Performs well on small datasets
-  - Produces stable and interpretable results
-
-- **Feature Engineering (if used by the model):**
-  - **Waist-to-Hip Ratio:** `Abdomen / Hip`
-  - **Body Index (BMI, imperial):** `703 × Weight / Height²`
+yaml
+Code kopieren
 
-Only engineered features that were present during training are computed in the app.
+Engineered features are calculated **only if required by the trained model**.
 
 ---
 
@@ -94,13 +79,12 @@ Only engineered features that were present during training are computed in the a
 
 | Measurement | Unit |
 |------------|------|
-| Height | **Centimeters (cm)** → converted to inches internally |
+| Height | **Centimeters (cm)** (converted internally to inches) |
 | Weight | **Pounds (lbs)** |
-| Circumferences | **Inches** |
-| Output Density | Unitless |
-| Output BodyFat | Percentage (%) |
+| Body circumferences | **Inches** |
+| Model output | **Density** (unitless) |
 
-⚠️ Correct units are critical for meaningful predictions.
+⚠️ Correct units are essential for meaningful predictions.
 
 ---
 
@@ -108,33 +92,44 @@ Only engineered features that were present during training are computed in the a
 
 ### 🔹 Single Prediction
 - Interactive form with realistic input ranges
-- Automatic unit conversion and feature engineering
-- Density prediction + body fat calculation
-- Warnings for implausible outputs
-- Display of aligned model inputs
+- Automatic unit conversion (cm → inches)
+- Feature engineering and alignment
+- Density prediction with plausibility warnings
+- Display of final model inputs
 
 ### 🔹 Batch Prediction (CSV)
-- Upload a CSV file with the required features
+- Upload CSV files with required feature columns
 - Automatic feature alignment
-- Download predictions as CSV
+- Download predictions as a CSV file
 
 ---
 
-## 🚫 Disclaimer
+## ⚠️ Important Notes on Predictions
+
+- Human body density typically lies within a **very narrow range (≈ 0.95 – 1.10)**  
+- Predictions outside this range may indicate:
+- Unrealistic input values
+- Inputs outside the training distribution
 
-This application is **not a medical device** and must not be used for health diagnosis or clinical decision-making.  
-Predictions are for **educational and demonstration purposes only**.
+The app provides **warnings** when predicted density values fall outside the typical physiological range.
 
 ---
 
 ## 🧪 Limitations
 
 - Small dataset size limits generalization
-- Density prediction errors propagate through the Siri equation
-- Linear models may oversimplify physiological relationships
-- Predictions outside the training distribution may be unreliable
+- Linear regression may oversimplify complex physiological relationships
+- Predictions may be unreliable for extreme or unrealistic input values
+- The model is not intended for medical or clinical use
 
-These limitations are **explicitly shown and explained** in the app.
+These limitations are **explicitly communicated in the app interface**.
+
+---
+
+## 🚫 Disclaimer
+
+This application is **not a medical device**.  
+All predictions are provided for **educational and demonstration purposes only** and must not be used for health diagnosis or treatment decisions.
 
 ---
 
@@ -148,7 +143,7 @@ These limitations are **explicitly shown and explained** in the app.
 
 ---
 
-## 📎 Files
+## 📎 Repository Structure
 
 ├── app.py
 ├── ridge_model.pkl
@@ -162,5 +157,7 @@ Code kopieren
 
 ## ✅ Conclusion
 
-This project faithfully follows the original competition objective by predicting **body density first** and then converting it into **body fat percentage** using a known physiological formula.  
-By combining transparency, proper unit handling, and clear warnings, the application demonstrates both the **power and limitations of machine learning in real-world regression problems**.
+This project demonstrates a complete and transparent machine learning pipeline for predicting **body density** using Ridge Regression.  
+By focusing directly on the competition target variable and avoiding unnecessary post-processing, the application provides **stable, interpretable, and scientifically consistent predictions**, making it well suited as a portfolio and learning project.
+
+---