Self-Organizing Maps and Evaluator Expertise for Transparent and Reproducible Consensus-Based Indicator Selection
Description
This dataset contains supplementary tables used in the analysis of evaluator expertise and indicator clustering with Self-Organizing Maps (SOMs). Each table reports the hyperparameter settings and performance metrics for SOM training, including: Initial and final sigma values Initial and final learning rates Initial, final, and minimum quantization error Number of neurons used Distribution of indicators across neurons The dataset supports the reproducibility of the clustering analysis presented in the manuscript Self-Organizing Maps and Evaluator Expertise for Transparent and Reproducible Consensus-Based Indicator Selection. Future versions of this dataset will include additional processed files and Jupyter notebooks used in the analysis.
Files
Steps to reproduce
1. Train Self-Organizing Maps using the hyperparameter settings provided in the tables (initial/final sigma, learning rate). 2. Track quantization error across training iterations. 3. Compare indicator distribution across neurons for different evaluator groups (high expertise, low expertise, full panel). 4. Use Adjusted Rand Index and dispersion metrics to assess clustering stability and agreement.