DSS-Group04

Running the App

To run the frontend:

1. Make sure Docker is installed and running. Refer to Docker

2. Download and unzip Group04Midterm.zip, then open the App folder

3. From the project root, run:

   docker compose up --build
   

4. Once the container is ready, go to localhost:8501 to view the dashboard.

Links:

• Codebase: GitHub
• Datasets:
  • RDW
  • REGIONAL
  • GEO

Data Collection and Preparation

A unified ETL pipeline integrates national open data from the RDW, CBS and KiM.
The Extract phase retrieves raw records and the Transform phase cleans, harmonises and standardises the data:

• RDW notebooks classify car models by fuel type, body type and weight (01_RDW_extraction.ipynb, 02_RDW_cleaning.ipynb, 03_RDW_classification.ipynb).
• Regional scripts merge PC6-PC4 areas and calculate proportions and Z-scores(01_REGIONAL_extr_transf.py, 02_NBH_REGIONAL_cleaning.py, 02_PC4_REGIONAL_cleaning.py, 02_REGIONAL_cleaning.py) .

The Load phase exports the final tables to BigQuery and populates the dashboard via Dockerised Streamlit components:

• Regional script: 03_REGIONAL_load.py .

Data Sources

• RDW Rijksdienst Wegverkeer Gekentekende Voertuigen (2025) link

• CBS Centraal Bureau voor de Statistiek

  • Kerncijfers Wijken en Buurten (2024) link
  • Kerncijfers Postcode 4 (2024) link
  • Buurt, Wijk en Gemeente (2023) voor Postcode + Huisnummer link
  • KiM Kennisinstituut Mobilitiet, Ministerie van Infrastructuur en Waterstaat. Atlas van de Auto (2024) link
  • Open Charge Map (OCM) Link
  • Google Custom Search API Link

Existing Indicators and Visualizations

1. Vehicle Search and Model Details

   • CSF/KPI: Data completeness, quality, and market dynamics.
   • Data Source: RDW vehicle dataset (RDW.rdw_classified, BigQuery).
   • Computation: Queries car specs (fuel type, mass, average price) directly; no derived KPI yet.
   • Code File: main.py (functions get_unique_brands(), get_car_details()).

2. Neighbourhood Map (Geo Layer)

   • CSF/KPI: Data consistency, look-alike matching (future).
   • Data Source: CBS GeoPackage 2023 (MAP.geo_neighbourhoods).
   • Computation: Loads and converts geometries to WGS84; color intensity currently random.
   • Code File: main.py (query_map_geometry() and PyDeck GeoJsonLayer).

3. Regional Compatibility – MatchScore

   • CSF/KPI: Car–Market Fit indicator measuring how well a car model aligns with regional socio-economic profiles.
   • Data Source: Combined RDW + CBS Regional data.
   • Computation: Weighted Manhattan distance between car attributes (body/fuel type, price) and regional indicators (income, fleet composition).
   • Visualization: Heatmap of PC4 regions and Top-N table with region name, score, affordability, and interest.
   • Code File: analysis_matchscore.py.

4. Market Positioning Indicators (Popularity, Niche)

   • CSF/KPI: Market dynamics and positioning within vehicle segments.
   • Data Source: RDW.
   • Visualization: Radar or bar charts showing model distribution across body classes.
   • Code File: analysis_marketposition.py.

5. Market Potential & Forecasting

   • CSF/KPI: Projected Market Growth and Forecast Accuracy (SMAPE).
   • Data Source: Dynamic RDW API (registrations 2020–2025).
   • Computation: Holt–Winters Exponential Smoothing with additive trend and seasonality. Forecasts displayed with ±80% prediction intervals.
   • Visualization: Line chart showing historical vs. predicted registrations for new and second-hand vehicles.
   • Code File: forecasting.py.

6. EV Readiness & What-If Scenarios

   • CSF/KPI: Electric Vehicle Readiness and Transition Potential.
   • Data Source: KiM Atlas van de Auto + Open Charge Map API (charger density, power level).
   • Computation: K-Means Clustering on standardized features (income, urbanization, fuel mix, household size).
     Includes Ridge Regression “what-if” model simulating EV adoption under changing socio-economic conditions.
   • Visualization: Clustered map of EV readiness, table of regional indicators, and interactive slider for scenario testing.
   • Code File: analysis_ev.py.

7. Vehicle Profile Explorer

   • CSF/KPI: Data completeness and quality check of RDW database.
   • Data Source: RDW classified dataset.
   • Computation: Retrieves model specs (mass, seats, propulsion, price) and compares them with segment averages.
   • Visualization: Model detail card with representative image (Google Custom Search API) and radar diagram summarizing vehicle attributes.
   • Code File: main.py.