← Back to The IndexFinal ReviewGroup 4
Final Review

Review Nº 04

Spatial Determinants of Data Center Location in Europe: A NUTS3-Level Empirical Analysis
30/30
Score
Strong and ambitious report with both classification and count modelling, but several modelling choices and interpretations need stronger validation and clearer separation between prediction and inference.
Data Centres · Final Review

The Pros

1 Item
+
Rich literature framing; clear hypotheses; uses both classification and regression/count perspectives; reports logistic regression significance, held-out classification metrics, and many regression alternatives; directly compares simple and complex models; includes discussion of policy implications and transatlantic comparison.

The Cons

1 Item
−
Regression becomes dominant even though the project explicitly asks for classification plus most accurate model; zero-imputation for renewable/IXP features may conflate missingness with absence; the best count model may be driven by extreme hubs; coefficient-based feature importance is overinterpreted; spatial/country dependence is not adequately handled.
↩ Back to The Index
Final Review · Group 4The IndexData Centres · 2026