I Built a Renewable Energy Siting Model for 500,000 Sites Across the US. Here's Everything I Learned.
I've listed geopandas on my resume for a while now.
Used it? Yes. Actually understood what was happening under the hood? ...mostly.
This week I decided to change that. No tuts. No documentation skimmi
aasthardg.hashnode.dev4 min read
The article presents a compelling geospatial AI/ML solution that combines Random Forest classification with Multi-Criteria Decision Analysis (MCDA) for renewable energy siting across the United States. The content effectively demonstrates the practical application of GIS, machine learning, and energy analytics to address a real-world sustainability challenge. The overall structure, methodology, and business relevance are well articulated. Excellent work. I do, however, have a few observations and questions that may help further strengthen the article:
Data Source Clarification: It is not entirely clear whether the analysis has been performed using real-world datasets or synthetically generated data. Providing additional clarity on the nature, source, and scope of the datasets used would enhance the credibility and reproducibility of the study.
Weighting Methodology: The final suitability score incorporates a weighting of 60% Random Forest output and 40% MCDA score. It would be beneficial to understand the rationale behind selecting this specific weighting scheme. Was this based on empirical validation, domain expertise, sensitivity analysis, or industry best practices?
Sensitivity Analysis: Have you evaluated alternative weighting combinations, such as 40% Random Forest and 60% MCDA, or other scenarios? It would be valuable to understand how sensitive the final rankings are to changes in these weights and whether the identified high-potential sites remain consistent across different weighting configurations.
Addressing these points would provide additional transparency into the methodology and further strengthen the technical rigor of an already impressive piece of work.
Overall, this is a well-executed and insightful article that showcases strong geospatial analytics and machine learning capabilities.