BEST: multi-scale territorial balance and urban well-being
Figure: indicator correlation heatmap for the BEST project (author, OECD Better Life Index data)

BEST: multi-scale territorial balance and urban well-being

Université de Technologie de Compiègne · Avenues laboratory (EA 7284) · TN09 final internship

PeriodSep 2024 - Feb 2025 My roleResearch assistant engineer InstitutionUTC · Avenues (EA 7284)

Repository ↗ Academic record →

Context

This was my TN09 final internship at UTC’s Avenues laboratory (EA 7284), contributing to BEST — a project building the theoretical foundation for “territorial system balance” and preparing a multi-scale decision-support tool for planners and policymakers.

Growing territories face four tensions: saturated transport, territorial inequality, rapid urbanisation and high housing costs. Existing models (urban dynamics, MCDA, PSR/DPSIR) treat housing, mobility, environment and economy separately; none offers a unified representation of the territorial system.

BEST aims to:

  • build a theoretical framework for “balance” applied to territorial systems;
  • characterise a dynamic, multi-scale territorial balance whose final criterion is residents’ well-being;
  • prepare the conceptual model, indicators and data for a future decision-support tool.

What I did

Three independent systematic reviews, an operationalisation of concepts into indicators, and a quantitative exploration of their relationships.

1. PRISMA systematic reviews

Searched Scopus, Google Scholar, HAL and PROSPERO, mainly in French and English, prioritising the last ten years, and managed references with Zotero.

The protocol followed PRISMA: identify studies → remove duplicates → screen titles and abstracts → full-text reading and inclusion criteria. Each article was classified as relevant / partially relevant / not relevant, with a justification.

ThemeArticles analysedGuiding question
Territorial system360How is a territory organised, and at which scales?
Well-being120Which dimensions make up residents’ well-being?
Territorial balance80Which levers support harmonious development?
Total560

2. Indicator frameworks

  • Territorial system: the 134 urban sustainability indicators of Agenda 21, classified by economic, social, environmental and institutional dimensions, supplemented with French sources (INSEE, ADEME, eaufrance, ATMO France, ministries).
  • Well-being: the 11 dimensions of the OECD Better Life Index — housing, income, jobs, community, education, environment, civic engagement, health, life satisfaction, safety, work-life balance — distinguishing objective (homicide rate) and subjective (feeling safe walking alone at night) indicators.

Each row documents the definition, official source and calculation method, so the indicators can be implemented directly.

3. Correlation analysis

The goal was to check the internal consistency of the indicator set and identify dependencies to include in the model.

OECD CSV (long format) → pivot country × indicator → Pearson matrix
                        → seaborn heatmaps → comparison by country group
Correlation heatmap of 22 OECD Better Life Index indicators. Most relationships match common sense; a few anomalies are kept as questions for later modelling.
Correlation heatmap of 22 OECD Better Life Index indicators. Most relationships match common sense; a few anomalies are kept as questions for later modelling.

Representative relationships:

RelationshiprReading
Life expectancy ↔ dwellings without basic facilities−0.85Poor sanitation undermines health
Personal earnings ↔ life satisfaction+0.80Perceived well-being is strongly linked to financial stability
Unemployment rate ↔ student skills−0.68Qualification and employability are linked

One anomaly was kept: water quality correlates strongly with employment and life satisfaction. It is more likely a sample-structure effect than a causal mechanism, and needs explanation in later modelling.

4. Conceptual modelling

Combined Forrester’s urban dynamics (population / housing / economic activity and feedback loops), MCDA, PSR / DPSIR and a sensitivity model into a multi-scale framework.

Tools: Vensim (system dynamics), ArcGIS (spatial analysis and mapping), Python (indicator processing), Excel (indicator tables).

Results

  1. A structured corpus of 560 articles and three conceptual reviews defining territorial systems, well-being and balance, and their links.
  2. Two ready-to-use indicator frameworks with official sources and formulas, providing the quantitative base for the future tool.
  3. A reproducible correlation analysis (code, data, figures) showing that most Better Life Index relationships match expectations, while isolating anomalies.
  4. A contribution to the BEST conceptual framework and decision-support algorithm.
TN09 academic poster: problem, territorial system, well-being, balance and the multi-scale framework.
TN09 academic poster: problem, territorial system, well-being, balance and the multi-scale framework.

Limits

  • The corpus is mainly English; French literature was used complementarily, not systematically.
  • Correlation is not causation: the analysis is exploratory, without significance testing or control for confounders.
  • Missing values were filled with 0 following the original analysis, which may bias some correlations for incomplete countries.
  • Indicator units, scales and periods remain heterogeneous; aggregating them into a single index was not the task of this internship.

Public files

01 TN09 internship report (PDF) 35 pages, including the three reviews, indicator frameworks, correlation analysis and conceptual model. 02 TN09 defence deck (PDF) Internship defence presentation. 03 TN09 academic poster (PDF) BEST project poster.

Technical notes

01 After 560 papers: turning territorial systems, well-being and balance into indicators PRISMA search, three systematic reviews, 134 sustainability indicators and the 11 dimensions of the OECD Better Life Index. 02 From a correlation heatmap to system dynamics: the BEST multi-scale framework 22 OECD indicators, Vensim system dynamics and ArcGIS spatialisation — and why correlation is not causation.

Related skills

  • Zotero
  • ArcGIS
  • Academic writing
  • Project management