Smoker Status Prediction — Flask Web App

A lightweight Flask application that predicts whether a person is a Smoker or Non-Smoker from biometric and clinical measurements.
Just fill in the form, pick a trained model, and get an instant colour-coded result.

The models were built in two Jupyter notebooks:

Notebook	Strategy
smokerStatus_v6-MANUAL_FeatureEng.ipynb	Domain-driven, manual feature engineering
smokerStatus_v7_MATH_FeatureEng.ipynb	Automatic additional feature creation using pairwise (add / multiply)

Selected exported pipelines reside in app/models/ and are loaded dynamically at runtime. Not all exported to app/models/ due to huge model file size

🎯 Course Context

This repository is submitted as Project #04 — IndonesiaAI_ML_Batch7_Project_04, the fourth official assignment in the Indonesia AI Machine Learning Batch 7 program.

---

✨ Key Features

Feature	Description
Interactive form	Collects 20 + biometric and sensory inputs such as blood pressure, cholesterol, liver enzymes, hearing & eyesight
Model picker	Choose among multiple Logistic-Regression, Random-Forest, and XGBoost pipelines trained on the same dataset
Sample-data autofill	“Fill form with sample data” loads a random row from static/data_contoh.csv for a one-click demo
Instant feedback	Prediction box turns green (Non-Smoker) or red (Smoker) immediately after submission
Open dataset	Uses the Playground Series – Season 3, Episode 24 health dataset

---

🗂️ Project Layout

app/
├── app.py                 # Flask entry-point
├── .flaskenv              # Local Flask settings (optional)
├── requirements.txt       # Python dependencies
├── templates/
│   └── index.html         # Single-page UI
├── static/
│   ├── style.css
│   ├── ai-01.png
│   └── data_contoh.csv    # Sample rows for autofill
├── models/                # Trained model folders (*.pkl + scaler + modelrun.py)
└── tree.py                # Utility script (decision-tree visualisation)

---

🚀 Quick Start

1. Clone this repository and open a terminal in its root folder (app/).

2. Create & activate a virtual environment (Python ≥ 3.8):

   python -m venv .venv
   # Windows
   .venv\Scripts\activate
   # macOS / Linux
   source .venv/bin/activate

3. Install dependencies:

   pip install -r requirements.txt

4. Run the server:

   # Option A – via Flask’s CLI (.flaskenv will enable debug by default)
   flask run
   # Option B – directly
   python app.py

5. Browse to http://127.0.0.1:5000 and start predicting!

---

🏗️ How It Works

1. Dynamic model loading
   On submission, app.py imports the chosen model’s modelrun.py on-the-fly and passes the form data to runflow() for prediction.

2. Sample-data simulation
   The /simulate route picks a random row from static/data_contoh.csv, removes the true smoking label, pre-selects a random model, and renders the form with those values.

3. Single-page UX
   All validation, radio buttons, and result styling are defined in templates/index.html.

---

📑 Training Notebooks & Dataset

File	Purpose
smokerStatus_v6-MANUAL_FeatureEng.ipynb	Manual feature engineering notebook
smokerStatus_v7_MATH_FeatureEng.ipynb	Automatic add/multiply feature engineering notebook

Not all model created by these two .ipynb exported to app/models/ due to huge model file size

Dataset: Playground Series – Season 3, Episode 24 (public health-check dataset for smoker-status classification).

---

Enjoy predicting — and feel free to fork, extend, or swap in your own models!

---

---

---

---

---

---

---

Smoker Status Prediction — Web Interface Manual

This web application allows users to predict smoking status (Smoker or Non-Smoker) based on biometric and clinical data input. The prediction is powered by machine learning models trained on health datasets. You can visit the online demo for a quick review.

Table of Contents

1. Input Form (No. 1)
2. Model Selector (No. 2)
3. Submit Button (No. 3)
4. Clear Button (No. 4)
5. Fill Sample Button (No. 5)
6. Prediction Result (No. 6)
7. Reference Info (No. 7)

---

1. Input Form

Location: Top section of the page

Enter biometric and medical data of the subject/patient.
These include:

• Age (in 5-year intervals)
• Height (cm)
• Weight (kg)
• Waist circumference (cm)
• Eyesight (Left & Right)
• Hearing (Left & Right)
• Blood pressure (Systolic & Diastolic)
• Fasting blood sugar (mg/dL)
• Cholesterol (Total, HDL, LDL, Triglyceride)
• Hemoglobin (g/dL)
• Urine protein level (0–4)
• Serum creatinine (mg/dL)
• Liver enzymes (AST, ALT, GTP)
• Dental caries (0 = no, 1 = yes)

Refer to the reference section below for valid value ranges.

---

2. Model Selector

Location: Dropdown below the form

Choose a trained machine learning model to perform prediction.
Available options include models like:

• Logistic Regression
• Random Forest
• XGBoost

---

3. Submit Button

Label: Submit

Click this to send the input data to the selected ML model.
The model will return a prediction of whether the subject is a Smoker or Non-Smoker.

---

4. Clear Button

Label: Clear

This button will clear all fields in the form.
Useful if you want to reset the form before entering new data.

---

5. Fill Sample Button

Label: Fill with Example

Click this to autofill the form with a random sample from the dataset.
This is helpful for testing the app or demoing predictions.
Note: It will exclude the actual smoking status during prediction.

---

6. Prediction Result

Location: Below the Submit button

• If the model predicts Non-Smoker, the result box will appear green.
• If the model predicts Smoker, it will appear red.

Use this prediction to guide further analysis or decision-making.

---

7. Reference Info

Located: At the bottom of the page

Provides:

• Dataset origin: Binary Prediction of Smoker Status using Bio-Signals | Kaggle
• Field descriptions: What each input represents and how it relates to smoking status

---

Notes

• Target: The goal is to predict the smoking status (0 = Non-Smoker, 1 = Smoker)
• Outlier Handling: Backend has safeguards for invalid inputs

---

Created as part of the Smoker Status Prediction Project — Machine Learning Batch 7