Bio-Signal Analysis for Smoking Prediction

Overview

This repository contains a machine learning project focused on predicting smoking habits using bio-signal data. The project aims to build a classification model that can accurately determine the presence or absence of smoking in an individual based on a dataset of biological and health-related attributes.

Problem Statement

The goal is to develop an intelligent system that can analyze a person's bio-signal data and classify whether they are a smoker or not. This can be valuable for global companies and research institutions looking to understand smoking patterns and develop targeted interventions.

Dataset

The dataset used in this project is a collection of basic health biological signal data, containing approximately 55,000 records with 27 attributes. The dataset includes features such as:

• ID: Index
• gender: Gender of a person (M or F)
• age: Age of a person (5-year gap)
• height(cm): Height of a person
• weight(kg): Weight of a person
• waist(cm): Waist circumference length
• eyesight(left): Left eyesight
• eyesight(right): Right eyesight
• hearing(left): Hearing pulse in left ear
• hearing(right): Hearing pulse in right ear
• systolic: Blood pressure (systolic)
• relaxation: Blood pressure (relaxation)
• fasting blood sugar: Blood test
• Cholesterol: Total cholesterol
• triglyceride: Lipid found in blood
• HDL: HDL cholesterol type
• LDL: LDL cholesterol type
• hemoglobin: Transporting oxygen in blood
• Urine protein: Excess of bloodborne proteins in urine
• serum creatinine: Amount of creatinine in blood
• AST: Glutamic oxaloacetic transaminase type
• ALT: Glutamic pyruvic transaminase type
• Gtp: γ-GTP
• oral: Oral Examination status
• dental caries: Tooth decay
• tartar: Tartar status
• smoking: Smoker (0 or 1) - Target Variable

Tasks Performed

The following tasks were performed in this project:

1. Data Loading:

   • Loading the data using Pandas.
   • Removing irrelevant columns (if any).
   • Checking the shape and datatypes of the dataframe.
   • Calculating statistical data (mean, std, min, max, etc.).

2. Data Cleaning:

   • Checking for missing values.
   • Handling missing values (if any).
   • Performing One-Hot Encoding for categorical features (gender).

3. Data Visualization:

   • Visualizing the distribution of smokers vs. non-smokers.
   • Analyzing smoking prevalence by gender.
   • Analyzing smoking prevalence by age group.
   • Detecting outliers using box plots (and deciding whether to remove or keep them based on domain knowledge).

4. Feature Selection:

   • Calculating feature importance using a suitable method (e.g., Random Forest, SelectKBest).
   • Selecting the top N features based on their importance scores.

5. Model Building & Evaluation:

   • Splitting the dataset into training and testing sets.
   • Building and evaluating the following classification models:
     • Logistic Regression
     • Decision Tree
     • Bagging Classifier
     • Extra Trees
     • Random Forest
   • Calculating accuracy and generating classification reports for each model.