WoonprijsML — Interactive Prediction Dashboard

Explore Noord-Holland

Click on any neighborhood polygon on the map or use the search bar above to inspect predictive real estate appreciation and satellite greenness/urban indices in Amsterdam, Haarlem, Haarlemmermeer, Heemstede, Hillegom, and Bloemendaal (including Bennebroek).

Region Highlights

Total Neighborhoods 724

Avg. 5Y Appreciation +48.2%

Model Engine Random Forest

ML Training Methodology

1. Multi-Period Panel Dataset

Instead of a single time-slice, the model was trained on a stacked panel dataset consisting of two 5-year observation cycles to capture temporal dynamics:
• Period A (2018 → 2023): 2018 baselines predicting 2023 prices.
• Period B (2020 → 2025): 2020 baselines predicting 2025 prices.
Combined, this builds a robust training base of 153 neighborhood records.

2. Biophysical & Socioeconomic Features

For every training period, features are derived from:
• WOZ Values (CBS): Base land value.
• Greenness & Built-up (NDVI/NDBI): Sentinel-2 summer indices.
• Surface Temp (LST): Landsat-8 land surface heat profiles.
• Nighttime Lights (VIIRS): Artificial radiance / economic activity.
• Crime Rate (CBS Dataderden): Registered crimes per 1,000 residents.
• Social & Infrastructure (CBS): Proximity to supermarkets, schools, daycares, average income, demographic share, and population density.

3. Model Selection & Evaluation

Multiple regression models (Linear, Ridge, Gradient Boosting, and Random Forest) are trained and compared. The pipeline dynamically selects the model with the highest generalization performance (R-squared on validation test partitions):
• Model Selected: Random Forest Regressor (Test R²: 0.440).
• Why Random Forest?: As an ensemble of decision trees, it effectively captures complex, non-linear interactions between spatial parameters (like ground temperature, demographics, and density) and real estate price trajectories, while mitigating overfitting compared to linear models or deep trees.

4. Feature Importances (Random Forest)

The trained Random Forest model utilizes the features with the following importances:
• Surface Temp (LST): 46.1%
• Average Income: 10.0%
• Demographics (Young Adults): 6.8%
• Crime Rate: 6.5%
• Built-up (NDBI) & Nightlights: 11.0%
• Other (Greenness/Distances): 19.6%

5. Model Data Matrix (Features Checklist)

Variable	Source	Hypothesis
NDVI (Greenness)	Sentinel-2	Lush vegetation boosts value.
LST (Surface Temp)	Landsat-8	Urban Heat Islands deter buyers.
NDBI (Built-up)	Sentinel-2	Tracks development density.
Nighttime Lights	VIIRS	Measures economic activity.
Supermarket Dist.	CBS	Shorter reflects convenience.
School / Daycare	CBS	Key driver for families.
Avg. Income	CBS	Gentrification index of area.
Young Adults %	CBS	Primary starter buyers cohort.
Pop. Density	CBS	Limits growth if over-crowded.
Crime Rate	CBS	Safety is prerequisite for value.

5. 2025 → 2030 Inference

Using the learned coefficients and the latest **2025 summer composite baselines**, the model predicts the relative growth rates expected through 2030.

Haarlemmermeer

Neighborhood Name

BU03940101

5-Year Predicted Appreciation

+84.2%

Estimated value increase from 2025 to 2030 based on biophysical and socioeconomic indicators.

Loading explanation...

Estimate Your Property Value

Your 2025 WOZ (€):

Projected 2030 Value: €0

Baseline WOZ (2025)€482,000
                                    Haarlemmermeer Avg: €460.000
                                    NL Avg: €370.000
                                

Greenness (NDVI)

0.45 Index

City Avg: 0.42 NL Avg: 0.45

Normalized Difference Vegetation Index: Measures vegetation density and tree cover. Green canopy space is highly attractive to buyers, exerting the strongest positive influence on our 5Y appreciation forecast model (+14.64 coefficient).

Built-up (NDBI)

-0.12 Index

City Avg: -0.11 NL Avg: -0.15

Normalized Difference Built-Up Index: Tracks density of buildings, runways, and paved roads. Controlled local infrastructure development supports demand, driving positive appreciation (+13.36 coefficient).

Surface Temp (LST)

24.8°C Celsius

City Avg: 28.2°C NL Avg: 26.5°C

Land Surface Temperature: Measures heat radiating from pavement, runways, and concrete during hot summer days (Urban Heat Island). Higher heat retention reduces comfort, resulting in a negative price influence (-2.97 coefficient).

Crime Rate (2025)

0.0 per 1k Inh.

City Avg: 14.5 NL Avg: 18.2

Crime Rate: Total registered crimes per 1,000 residents in the neighborhood. Calculated using police reports from CBS Dataderden and normalized by local population. Lower rates indicate safer neighborhoods.

Nighttime Lights

0.0 nW/cm²/sr

City Avg: 12.8 NL Avg: 9.5

VIIRS Nighttime Radiance: Measures average artificial light emitted at night. Higher values indicate concentration of commercial infrastructure, street lighting, and general economic activity.

Average Income

€0.0 per Inh.

City Avg: €32,400 NL Avg: €31,100

Average Income: Mean annual income per resident in the neighborhood (in EUR × 1000). A high or growing income profile is a primary indicator of neighborhood gentrification and local purchasing power.

Young Adults Share

0% of Pop.

City Avg: 24.8% NL Avg: 26.2%

Age Group 25-45 Share: Percentage of the local population aged 25 to 45 years. This demographic cohort contains young professionals and starter families who drive the bulk of residential real estate purchases.

Population Density

0 Inh. / km²

City Avg: 2,100 NL Avg: 530

Population Density: Number of inhabitants per square kilometer. Balanced density indicates active urban centers; extremely high density without sufficient green buffers can limit residential price growth potential.

Supermarket Dist.

0.0 km

City Avg: 0.8 km NL Avg: 0.9 km

Distance to Supermarket: Average walking/driving distance in kilometers to the nearest large supermarket. Shorter distances reflect higher baseline daily residential convenience.

School Distance

0.0 km

City Avg: 0.6 km NL Avg: 0.7 km

Distance to Primary School: Average distance in kilometers to the nearest primary school. Proximity to basic education facilities is a primary priority for family home buyers.

Daycare Distance

0.0 km

City Avg: 0.5 km NL Avg: 0.6 km

Distance to Daycare: Average distance in kilometers to the nearest professional daycare / kinderdagverblijf facility, essential for working parents.

Predicted 5Y Growth

0% 50% 100%