Diarrhea-Prediction-Model

Project Overview

This project analyzes and predicts diarrhea cases using historical data from four divisions in Bangladesh: Rajshahi, Khulna, Dhaka, and Chattogram. The goal is to explore trends, identify significant correlations with weather variables, and create predictive models for better disease management.

Features

• Data Exploration: Examines trends in diarrhea cases across regions and their relationship with weather variables like temperature, humidity, and precipitation.
• Data Cleaning: Implements robust methods to handle outliers while preserving seasonal and trend components.
• Time Series Analysis: Decomposes diarrhea cases into seasonal, trend, and residual components for better understanding.
• Machine Learning Models: Predicts diarrhea cases using Linear Regression, Random Forest, Support Vector Regression (SVR), and Decision Tree.
• Visualization: Provides visual insights through line plots, heatmaps, and decomposition graphs.

Data Sources

• Regions: Rajshahi, Khulna, Dhaka, and Chattogram.
• Variables: Diarrhea cases, minimum/maximum temperature, humidity, and precipitation.

Installation

1. Clone the repository:

   git clone https://github.com/your-repo/diarrhea-prediction.git
   cd diarrhea-prediction

2. Install required Python packages:

   pip install -r requirements.txt

Usage

1. Load datasets:
   • Update data paths in diarrhea_analysis.py or use the linked GitHub source.
2. Run the script to perform:
   • Data preprocessing and cleaning.
   • Exploratory data analysis (EDA).
   • Time series decomposition.
   • Machine learning model training and evaluation.
3. Visualizations will be generated and displayed in interactive plots.

Workflow

1. Data Preprocessing:

   • Combines datasets for all divisions.
   • Handles missing values and performs outlier detection.

2. Exploratory Data Analysis (EDA):

   • Identifies trends in diarrhea cases across regions.
   • Examines correlations between weather variables and diarrhea cases.

3. Time Series Analysis:

   • Decomposes time series into seasonal, trend, and residual components.

4. Machine Learning:

   • Trains predictive models using weather variables and trends.

5. Visualization:

   • Generates plots to compare actual vs. predicted values.

Results

• Minimum temperature showed the strongest correlation with diarrhea cases.
• Random Forest outperformed other models in accuracy and predictive power.

Models

Model	Performance (R²)
Linear Regression	0.8
Random Forest	0.8
Support Vector Regression	0.7
Decision Tree	0.6

Contributions

Feel free to contribute by submitting pull requests for enhancements or bug fixes. Suggestions for new features and improvements are welcome!