---
id: "houses"
slug: "houses"
name: "Houses"
domain: "Real Estate & Urban Planning"
taskType: "regression"
numRows: 20640
numColumns: 9
targetColumn: "median_house_value_usd"
primaryMetric: "RMSE"
primaryScore: "40,971"
parquetUrl: "https://huggingface.co/datasets/carlahq/demo-tabular-benchmarks/resolve/main/data/houses/houses.parquet"
spreadsheetUrl: "https://docs.google.com/spreadsheets/d/1hMomH4Jicrd9_WTms7QI3orWGcKcX8Uo4hF0CuQjdjI/edit?usp=sharing"
canonicalUrl: "https://carlahq.eu/datasets/houses"
---

# Dataset: Houses (houses)

> The California Houses dataset contains 20,640 census block group records from the 1990 U.S. Census, structured to predict median house values using demographic, spatial, and structural real estate features.

## Dataset Overview

| Attribute | Value |
| :--- | :--- |
| **Task Type** | `regression` |
| **Business Domain** | Real Estate & Urban Planning |
| **Target Column** | `median_house_value_usd` |
| **Rows (Instances)** | 20,640 |
| **Features (Columns)** | 9 |
| **Primary Evaluation Metric** | RMSE |
| **Best Benchmark Score** | **40,971** |

## Business Decision Context
**Context:** Automated valuation models (AVMs) and property investment firms require granular, location-sensitive appraisals to assess neighborhood residential asset values accurately without manual site inspections.
**Objective:** Predict the median residential house value for geographic block groups based on spatial coordinates, household demographics, and structural housing metrics.
**Decision Impact:** Accurate property valuation mitigates underwriting risk in residential mortgage lending, eliminates appraisal latency, and prevents severe capital misallocation stemming from under- or over-valued real estate portfolios.

## Feature Schema
| Column Name | Data Type | Role | Distinct Values | Missing Values (%) | Description |
| :--- | :--- | :--- | :--- | :--- | :--- |
| `median_house_value_usd` | `numeric` | target | 3,842 | 0 | Median house value for households within the census block group |
| `median_income_tens_of_thousands` | `numeric` | feature | 12,928 | 0 | Median household income within the census block group (expressed in tens of thousands of USD) |
| `housing_median_age_years` | `numeric` | feature | 52 | 0 | Median age of residential buildings within the census block group |
| `total_rooms_count` | `numeric` | feature | 5,926 | 0 | Total count of rooms across all residential units in the block group |
| `total_bedrooms_count` | `numeric` | feature | 1,928 | 0 | Total count of bedrooms across all residential units in the block group |
| `population_count` | `numeric` | feature | 3,888 | 0 | Total number of permanent residents residing within the census block group |
| `households_count` | `numeric` | feature | 1,815 | 0 | Total number of households (occupying housing units) in the block group |
| `latitude_deg` | `numeric` | feature | 862 | 0 | Centroid latitude coordinate of the census block group |
| `longitude_deg` | `numeric` | feature | 844 | 0 | Centroid longitude coordinate of the census block group |

## Empirical Benchmark Leaderboard
| Rank | Algorithm / Model | Model Family | Runtime | R² Score | MAE | RMSE |
| :--- | :--- | :--- | :--- | :--- | :--- | :--- |
| 1 | **Google TabFM v1.0** | Tabular Foundation Model | PyTorch/Python | **0.8944** | 22,351.1 | 37,481 |
| 2 | **TabPFN v3** | Tabular Foundation Model | PyTorch/Python | **0.8826** | 23,749.7 | 39,522.7 |
| 3 | **EXAONE Tabular** | Tabular Foundation Model | PyTorch/Python | **0.8800** | 24,044.7 | 39,970.9 |
| 4 | **Full TabICLv2 (PyTorch Reference)** | Tabular Foundation Model | PyTorch/Python | **0.8739** | 24,947.3 | 40,970.9 |
| 5 | **nanotabicl Vanilla** | Tabular Foundation Model | PyTorch/Python | **0.8737** | 24,965.6 | 41,005.3 |
| 6 | **Streaming nanotabicl** | Tabular Foundation Model | PyTorch/Python | **0.8717** | 25,242.7 | 41,314.9 |
| 7 | **Carla Engine (TabICLv2 WebGPU)** | Tabular Foundation Model | Browser | **0.8661** | 25,992.2 | 42,213.4 |

### Repeated Cross-Validation Fold Scores (TabICLv2 Reference)
| Fold | Train Samples | Test Samples | Metric Score | Latency |
| :--- | :--- | :--- | :--- | :--- |
| Fold 0 | 13760 | 6880 | 40329.2298 | 5045ms |
| Fold 1 | 13760 | 6880 | 41270.1006 | 5019ms |
| Fold 2 | 13760 | 6880 | 41615.9357 | 5059ms |
| Fold 0 | 13760 | 6880 | 40698.9324 | 5060ms |
| Fold 1 | 13760 | 6880 | 40859.5327 | 5061ms |
| Fold 2 | 13760 | 6880 | 40216.1469 | 5007ms |
| Fold 0 | 13760 | 6880 | 39977.5329 | 4987ms |
| Fold 1 | 13760 | 6880 | 41478.6979 | 5057ms |
| Fold 2 | 13760 | 6880 | 42291.9349 | 5074ms |

## Provenance & Downloadable Artifacts
- **Dataset Parquet File:** [Direct Parquet Download](https://huggingface.co/datasets/carlahq/demo-tabular-benchmarks/resolve/main/data/houses/houses.parquet)
- **Google Sheets Interactive Demo:** [Open in Google Sheets](https://docs.google.com/spreadsheets/d/1hMomH4Jicrd9_WTms7QI3orWGcKcX8Uo4hF0CuQjdjI/edit?usp=sharing)
- **OpenML Source:** [OpenML Dataset](https://www.openml.org/d/537)
- **Academic Paper / Citation:** [Read Research Paper](https://doi.org/10.1016/S0167-7152(96)00140-X)
- **Canonical HTML Page:** [https://carlahq.eu/datasets/houses](https://carlahq.eu/datasets/houses)