How Dometr Works
Dometr uses machine learning algorithms to analyze real estate listings and provide accurate valuations.
Data Collection
Dometr continuously collects data from major Serbian real estate portals, including:
- Property listings (apartments, houses, land)
- Property characteristics (size, rooms, floor, condition)
- Location data and coordinates
- Price history and market trends
- Infrastructure around properties (schools, shops, transport)
Machine Learning Models
Dometr uses several ML models:
- Price Prediction Model: Estimates fair market price based on property features, location, and comparable sales
- Rental Price Model: Predicts rental income based on property characteristics and market rates
- Yield Calculator: Calculates investment return (annual rental income / purchase price × 100)
- Luxury Level Classifier: Identifies luxury properties based on photos, location, and features
Data Imputation
For properties with missing data, Dometr uses statistical methods to estimate:
- Missing property characteristics
- Incomplete location information
- Unspecified infrastructure features
Valuation Process
When you search for a property, Dometr:
- Finds similar properties in the area (within 2km radius)
- Compares prices within ±20% range
- Applies ML models to calculate fair market price
- Calculates rental yield if applicable
- Identifies if the property is overpriced or underpriced
Accuracy
Dometr's accuracy depends on:
- Amount of market data available
- Quality of property listings
- Market conditions and trends
The service provides transparent calculations and shows comparisons with similar properties so you can make informed decisions.