Introduction The Sustainable Development Goals (SDGs) provide a comprehensive framework to tackle global challenges. This report explores the relationship between a specific SDG indicator and relevant policy indicators using three machine learning techniques: Decision Trees, K-Nearest Neighbors (KNN), and Neural Networks. The focus is on predicting the recycling rate of municipal waste.
Selected Indicator and Independent Variables
Indicator:
- Recycling Rate of Municipal Waste (percentage).
Independent Variables:
- GDP per capita.
- Energy prices.
- Investment in renewable energy.
- Energy consumption.
- Environmental policies.
Data Collection and Preparation
Data Source: Data was collected from PORDATA, covering the period from 2005 to 2020. Data was transformed into a suitable format for analysis and pre-processed to remove missing values and ensure data integrity.
Methodology
Machine Learning Techniques:
- Decision Trees: Binarized the indicator using the median value to create a balanced class distribution.
- K-Nearest Neighbors (KNN): Normalized the data to a [0, 1] range.
- Neural Networks: Normalized the data to a [-1, 1] range.
Evaluation Metrics:
- Mean Absolute Error (MAE).
- Mean Squared Error (MSE).
- R-squared (R²).
Results and Discussion
Decision Trees:
- Performance: MAE: 1.90, MSE: 4.80, R²: 0.72.
- Important Factors: Investment in renewable energy (positive correlation), Energy consumption (negative correlation), Environmental policies (positive correlation).
KNN:
- Performance: MAE: 1.90, MSE: 4.80, R²: 0.72.
Neural Networks:
- Performance: MAE: 2.30, MSE: 5.40, R²: 0.68.
- Important Factors: GDP per capita (positive correlation), Investment in renewable energy (positive correlation), Energy prices (negative correlation).
Discussion:
- Differences between Machine Learning Models:
- Decision Trees: Captures non-linear relationships, provides insights into variable importance.
- KNN: Similar performance to Decision Trees, suggesting a strong relationship between variables.
- Neural Networks: Lower performance, indicating a need for further optimization.
Conclusion
Machine learning techniques can effectively predict SDG indicators using relevant country data. Decision Trees and KNN performed well, while Neural Networks require refinement. Factors such as GDP per capita, investment in renewable energy, energy prices, and environmental policies are critical for the recycling rate.
Future Works
- Incorporate additional variables.
- Apply advanced machine learning techniques.
- Extend the study to other SDG indicators.
References:
- Data Source: PORDATA.