Artificial Intelligence (AI) Modeling Software

Vendors to provide a secure, enterprise-grade Artificial Intelligence (AI) modeling software platform designed specifically for property tax mass appraisal, capable of handling high-volume
spatial and characteristics data to improve annual valuation accuracy, productivity, and transparency.
- Required Components:
1. Parcel-level explainability
• The solution shall provide a parcel-level explanation of each modeled value, including the source data, variables used, material adjustments, applicable market area or neighborhood, model version, comparable properties when applicable, and separate land and improvement value conclusions.
2. Transparent and auditable modeling
• The solution shall not operate exclusively as a proprietary black box. Authorized County personnel must be able to review model specifications, variables, transformations, calibration results, performance measures, and the derivation of individual parcel values
3. Land and improvement modeling
• The solution shall independently model and report land and improvement components and shall support residential improved property, commercial improved property, and vacant land.
4. Multiple modeling methodologies
• The solution shall support multiple statistical and machine-learning methodologies and permit comparison or combination of model results. Proposers shall describe ensemble methodology, model-selection criteria, weighting, validation, and safeguards against overfitting.
5. Automated data-quality assessment
• The solution shall evaluate data quality at the parcel and field level, identify missing, inconsistent, anomalous, or potentially inaccurate attributes, and summarize findings by property class, geographic area, neighborhood, and responsible workflow.
6. Geospatial enrichment
• The solution shall derive relevant parcel characteristics from County and external geospatial data, including, where available, frontage, parcel geometry, topography, flood hazard, zoning, water influence, corner influence, cul-de-sac influence, accessibility, and other locational characteristics
7. Market-based neighborhood analysis
• The solution shall evaluate existing appraisal neighborhoods and recommend alternative boundaries based upon physical barriers, location, property characteristics, sales behavior, and statistically measurable market differences.
8. RealWare and GIS interoperability
• The solution shall demonstrate repeatable import from and export to the County’s Harris Govern RealWare environment and County GIS without requiring replacement of either system. The proposer shall document field mapping, validation, exception handling, reconciliation, update frequency, and protection of the system of record.
9. IAAO-compliant ratio studies
• The solution shall perform ratio studies consistent with current IAAO standards and calculate, at minimum, median assessment ratio, coefficient of dispersion, pricerelated differential, price-related bias, vertical-equity measures, confidence intervals where applicable, distribution graphics, scatterplots, and parcel-level identification of outliers.