Proprietary Technology · Spatial AI & Big Data IP reserved

Multi-Factor Valuation Model & Spatial Artificial Intelligence

The system integrates millions of real market data points, normalizes them across multiple dimensions and applies micro-spatial machine learning to determine the true value of a property objectively, scientifically and transparently.

The 5-stage data processing pipeline

Step 01

Multi-tier Collection

Real transactions, appraisals, market data & State price benchmarks

Step 02

AI Noise Removal

Filters out fake, inflated and undercut listings and outliers

Step 03

7-Dimension Normalization

Scale, shape, road type, legal status, negotiation...

Step 04

Dynamic Weighting

Micro-distance decay & time-decay cycle

Step 05

Conclusion & Confidence

Aggregates the true value and auto-scores confidence

Intellectual property notice: The core mathematical model, detailed weights, data structures and machine-learning algorithms are the exclusive intellectual property of DatNgop. This document gives an overview of the scientific methodology and the technical factors considered.

Part 01

Multi-tier Data Foundation & the Evidence Principle

DatNgop's valuation model is built on Evidence-Based Valuation, following the core principles of the International Valuation Standards (IVS) and adapted to the dynamics of the Vietnamese real estate market.

To ensure accuracy and avoid compounded errors, the system independently separates the value of land-use rights from the value of the buildings on the land:

1. Verified Transaction Data

Actual transfer transactions verified through field records, completely excluding figures that were declared inaccurately.

2. Independent Appraisal Records

Survey results and in-depth appraisal certificates from valuation experts, banks and reputable financial institutions.

3. Wide-coverage Market Data

Listing and asking-price data collected continuously in real time, with automatic price-structure extraction and multi-layer cleaning.

4. State Benchmark Network

State land price tables and local zoning plans at each location and street, serving as the legal reference baseline.

Principle of objectivity: If a property includes a house, the system never simply divides the total asking price by the area. Every comparable is separated and independently normalized to a pure land unit price before entering the comparison algorithm.

Part 02

The 7 Groups of In-depth Normalization Variables

In reality, no two properties are exactly alike. To compare them fairly and accurately, the system applies Multi-Variate Normalization to bring every property to a Standardized Baseline Price before comparison.

Location & Micro-spatial Proximity

Micro-spatial

GPS coordinates accurate to the meter, direct access distance, micro geographic zoning and connectivity to the planned core area.

Parcel Scale & Morphology (Scale Economics)

Morphology

Land area is adjusted by the law of diminishing marginal returns to scale (a larger parcel has a lower price per m² than a smaller one in the same location), combined with frontage ratio, depth and parcel shape (flaring, narrowing, square, diagonal corner, irregular).

Transport Infrastructure & Accessibility

Transport infrastructure

Road classification (main frontage, internal road, car alley, three-wheeler alley, motorbike alley), planned right-of-way, width of the road in front, through-road vs dead-end alley, and distance to the main arteries.

Legal Status & Land Zoning

Legal & zoning

Detailed land-use classification (urban residential ODT, rural residential ONT, perennial crops CLN, other annual crops HNK, commercial-service TMD...), land-use certificate status and term of use.

Market Negotiation Dynamics

Market behavior

A model that automatically estimates the gap between the public asking price and the expected transaction price for each segment and data type.

Commercial Potential & Utility

Commercial use

Ability to generate business or rental income, the maximum permitted building density and height, and how busy the street is.

Terrain Condition & Micro-climate

Micro condition

Approach direction, slope, drainage and seasonal flooding, feng shui position and the surrounding landscape.

Part 03

Noise Removal & Intelligent Outlier Rejection (AI Anomaly Detection)

The real estate market always contains noise: fake listings that chase views, inflated asking prices from speculators, undercut listings meant to snap up properties, and one-off family transactions that do not reflect the market.

DatNgop applies a Multi-Stage Outlier Rejection model:

Layer 1

State Price Band Matrix

Compares the standardized unit price with the State price band in the area to remove economically implausible data points.

Layer 2

Market Statistical Percentiles

Computes a preliminary weighted median and removes points that deviate beyond the standard statistical range of the cluster.

Layer 3

Optimal Weight Convergence

Re-estimates the optimal median from the cleaned sample, eliminating extreme skew.

User protection: Any data point showing signs of anomaly is automatically excluded from the valuation, so the result always reflects true transaction values in the market.

Part 04

Dynamic Spatiotemporal Weighting (Spatiotemporal Decay)

Real estate value is extremely local and moves with time cycles. Comparables are therefore not treated equally but given a Composite Weight by a spatial and temporal decay function:

Distance Decay

The nearest micro radius around the pin is prioritized. The closer a comparable, the more it contributes to the valuation, capturing the local price level accurately.

Time Decay

The most recent transactions in recent months are prioritized. Older data is automatically discounted by a decay function matching the market cycle.

Data Source Hierarchy

Properties with in-depth appraisals and field-verified transactions carry the highest credibility weight, followed by approved market data sources.

Part 05

Adaptive Cascade Architecture

So the system can value accurately in both busy cities and data-sparse areas across the country, the algorithm is structured in 3 Hierarchical Adaptive Levels:

Level 1 (UT1)

Micro-Cluster Direct Matching

Uses the comparable cluster within the nearest micro radius that shares the same land type and segment as the subject property. If this set meets the statistical confidence standard, the system concludes at this level.

Level 2 (UT2)

Cross-Segment Calibration

When same-type data is not dense enough, the system expands to nearby properties of other land types that can be converted, applying correlation conversion factors to add reliable comparables.

Level 3 (UT3)

Regional Benchmark Mapping

Used when the micro area has very sparse transactions. Instead of assigning far-away properties directly, the algorithm maps market correlations to nearby benchmark areas with similar economic and infrastructure characteristics.

Strict geographic principle: Far-away properties are never compared directly with nearby ones without a market-correlation normalization model, as this would distort the local price level.

Part 06

Confidence Score & Report Transparency

Every valuation report on DatNgop carries a Confidence Score, computed automatically from 4 quantitative criteria:

Available Data Density

The amount of evidence and transactions collected within the micro radius.

Spatial Proximity

The average distance from the comparables to the appraisal pin.

Time Recency

How recent the comparable transactions are within the latest cycle.

Statistical Consistency

The dispersion and convergence of the normalized market evidence.

The report also lists the evidence and real reference points in the area so users can verify them, keeping the input data transparent while protecting the proprietary core algorithm.

Part 07

Commitment to Objectivity & Legal Limits

• Absolute objectivity: The model runs 100% on automated algorithms, free from subjective influence or commercial bias from buyers, sellers or any broker.

• Intended use: Valuation results on DatNgop.com are reliable quantitative reference data for research, market surveys, investment decisions or preparing financial documents.

• Legal notice: Results do not replace an official valuation certificate under Vietnam's Law on Prices, which is issued by licensed appraisers after an on-site inspection and in-depth legal review.