Simple Local Regression for Mortality Rate Prediction

This project explores the use of local linear regression to predict mortality rates (MORT) based on the percentage of families living in poverty (POOR). A short report is published in the zenodo platform The dataset can be found in the file "pollution_cleaneddata.csv" with the code in the notebook file "Exercise_LinearRegression.ipynb".

The regression is formulated as a weighted least-squares optimization problem, where nearby observations are given different weights according to their distance from the prediction point.

The project compares three weighting methods and investigates how the choice of the number of neighboring observations affects the fitted regression and prediction accuracy. The models are used to estimate mortality rates for poverty levels of 10%, 18%, and 25%, including the corresponding standard errors.

The repository contains the implementation, visualizations, and results used to evaluate the different weighting strategies and neighborhood sizes.

Environment Setup

This project uses Conda for environment and package management. The required packages and versions are specified in environment.yml.

1. Create the environment

Clone the repository and navigate to the project directory:

git clone <repository-url>
cd <repository-name>

Create the Conda environment using the provided environment.yml file:

conda env create -f environment.yml

2. Activate the environment

conda activate statistics

FAIR Principles

This project follows the FAIR principles: Findable, Accessible, Interoperable, and Reusable.

• Findable: The project includes relevant keywords and a description of its purpose. The notebook file explains the code and the steps involved in the analysis. The project also has a Digital Object Identifier (DOI), providing a persistent identifier that makes it easier to find and cite.
• Accessible: The project and its datasets is available in the repository and can be downloaded by anyone. The project is licensed under the MIT License, and a environment.yml file is provided to help users install the necessary dependencies.
• Interoperable: The data is stored in CSV format, which is widely supported by programming languages and data analysis tools. A JSON file is also included to provide structured, machine-readable metadata, improving compatibility with other systems and tools.
• Reusable: The MIT License permits reuse under its terms. The explanatory notebooks, public dataset, and listed software dependencies help others understand, reproduce, and build upon the project. The DOI also supports proper citation and attribution.