Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/holy-angel-university/life-expectancy-prediction

The project aims to analyze and predict life expectancy at birth, which reflects the average number of years a newborn infant would live if current mortality trends persist throughout their lifetime.
https://github.com/holy-angel-university/life-expectancy-prediction

machine-learning python3 student-project

Last synced: about 1 month ago
JSON representation

The project aims to analyze and predict life expectancy at birth, which reflects the average number of years a newborn infant would live if current mortality trends persist throughout their lifetime.

Awesome Lists containing this project

README 

        
# Life Expectancy Prediction



The project aims to analyze and predict life expectancy at birth, which reflects the average number of years a newborn infant would live if current mortality trends persist throughout their lifetime. By utilizing a variety of data science techniques, including Linear Regression, Decision Trees, Random Forest, and Support Vector Machines (SVM), the project will explore historical life expectancy data derived from multiple authoritative sources such as the United Nations Population Division, Eurostat, and national statistical offices.



## About the Dataset

The dataset was obtained from [https://fred.stlouisfed.org/series/SPDYNLE00INUSA]. It has 2 Columns, one is the date born and the other is the number of year lived. The data is derived from male and female life expectancy at birth from sources such as, United Nations Population Division, World Population Prospects, United Nations Statistical Division, Population and Vital Statistics Report (various years), Census reports and other statistical publications from national statistical offices, Eurostat: Demographic Statistics, Secretariat of the Pacific Community: Statistics and Demography Program, and U.S. Census Bureau: International Database.



Note: The Month and Day of the dataset are all January 1 for a more accurate comparison, the only difference that they have is the year born



## Dataset Snapshot

The following table shows a snapshot of the dataset used in this project:

 -Date: Represents the date born.

 -SPDYNLE00INUSA: Represents the life expectancy.



## Summary of Findings

As the years past, life expectancy is slowly getting lower. There might be some years that life expectancy increase but there is a continues trend of it going down.



## Data Preprocessing

In the Data Preprocessing first the data was checked for null values, on the Year column it was changed to only the year since the values of the month and day are the same. Which is why only the year is reflected in the Year column.



## Exploratory Data Analysis

The following Graphs are based from the given dataset:



![alt text](assets/graph.png)

Here above you can see that in the early years life expectancy was slowly rising with a few minor dips, but in the recent years life expectancy had a major decline but is now slowly increasing again.



![alt text](assets/Matrix.png)

Here in the correlation matrix it is seen that the year you are born there is a huge correlation to your life expectancy. Which can be attributed to the customs and lifestyle of that year which affects the life span of a person.



![alt text](assets/Distribution.png)

Here is the distribution count of the number of person their maximum life expectancy from the dataset.



Below is the graph based from the prediction of the model:



![alt text](assets/Prediction.png)



Above it is predicted that life expectancy will increase for a few years but then will suffer a major decline, this shows that the current lifestyles of the people in the 21st century is leading to a lower life expectancy.



## Model Development

To predict life expectancy, we utelized four models: Linear Regression (with polynomial features), Decision Tree, Random Forest, and Support Vector Regressor. The data was standardized, and models were trained to capture life expectancy trends. Performance was evaluated using MSE and R² metrics to compare each model’s predictive accuracy. This approach ensured a balance between model complexity and accuracy in capturing the trend.



## Model Evaluation

The table below shows the evaluation of the models



| Model Name            |   MSE  |   R2   |

| :-------------------- | -------|--------|

| Linear Regression     | 0.9611 | 0.9018 |  

| Decision Tree         | 0.4162 | 0.9575 |

| Random Forest         | 0.3896 | 0.9602 |

| Support Vector Machine| 0.4085 | 0.9583 |



## Conclusion

This Life Expectancy Prediction project demonstrates the process of building predictive models to forecast life expectancy based on historical data. By carefully cleaning and preparing the data, we ensured that the models received high-quality input for training. Key insights into the dataset were drawn through exploratory visualizations, including trend analysis, correlation heatmaps, and target variable distributions. 



Multiple regression models, including Linear Regression, Decision Tree, Random Forest, and Support Vector Machine (SVM), were trained and evaluated based on Mean Squared Error (MSE) and R-squared (R2) scores. Each model contributed unique perspectives on prediction accuracy, with the best model selected based on its performance in accurately predicting life expectancy. The prediction results, along with the visualized trends, indicate that life expectancy can be reliably forecasted using machine learning models, especially when tuned to capture nonlinear patterns through polynomial features.



This project highlights the effectiveness of machine learning in predicting future values in a time series, such as life expectancy, with opportunities for future enhancements through feature expansion or more complex modeling techniques. These insights and predictions can aid public health planning, resource allocation, and policy-making efforts in addressing life expectancy trends.