monthly housing cost

Overview

This model estimates a voter’s monthly housing payment as a continuous dollar value.

Built as a boosted tree regressor on survey data, it integrates property records, consumer residence indicators, voter-file features, and area-level rent and home-value context to produce monthly cost estimates.

Organizations can use these scores to support threshold segmentation and analysis of housing cost distributions across geographies.

  • approach: dnn regressor
  • training rows: 5,379
  • last updated: 2026-09-06 18:24:02

Validation

metric value
R² (correlation) 0.30
root mean squared error 863
mean absolute error 661
huber loss 660
gini coefficient —

Lift Chart

Each bar shows how much more common the outcome is within a score decile than in the population overall. Bars taller than the dashed line (lift of 1) mean the model’s top-ranked groups concentrate the outcome; a steady decline from decile 1 to 10 indicates the model is ranking effectively.

Calibration Plot

Bars of similar height within each decile indicate well-calibrated predictions. Systematic differences reveal where the model over- or under-predicts.

Cumulative Gains

The further the model curve bows above the diagonal, the better the model concentrates the outcome in its top-ranked predictions.

Decile Analysis Table

decile count mean actual mean predicted mae lift
10 138 2,361 2,345 16 1.93
9 137 1,735 1,840 105 1.42
8 137 1,569 1,594 24 1.28
7 138 1,562 1,392 169 1.27
6 137 1,193 1,210 17 0.97
5 137 1,089 1,051 38 0.89
4 138 940 928 12 0.77
3 137 812 799 13 0.66
2 137 578 650 72 0.47
1 138 415 446 31 0.34

Architecture

This model is a deep neural network regressor trained with BigQuery ML.

Hyperparameter configuration

  • optimizer: adam
  • max iterations: 20
  • learn rate: 0.0029
  • l1 regularization: 0
  • l2 regularization: 0
  • min relative progress: 0.01
  • warm start: FALSE
  • early stop: TRUE
  • input label columns: affordability_monthly_housing_cost
  • data split method: AUTO_SPLIT
  • hidden units: 64, 32, 16
  • batch size: 64
  • dropout: 0
  • num trials: 18
  • max parallel trials: 5
  • hparam tuning objectives: R_SQUARED
  • enable global explain: FALSE
  • tf version: 1.15
  • activation fn: elu

Features

Importance

Sourcing

Agency for Healthcare Research and Quality

  • tract % direct-purchase insurance only
  • tract % owner costs 30%+ of income

Attom Property Data API

  • last sale price
  • property tax billed

Revelio Labs

  • avg tenure (days)

State voter files (via TargetSmart)

  • age
  • years registered to vote

TargetSmart

  • child tax credit support score
  • children present score
  • homeowner score
  • income rank
  • off-year general turnout score
  • paid leave support score
  • race: white

U.S. Census American Community Survey

  • block group median rent

Verisk (via TargetSmart)

  • home value
  • income
  • residence tenure
  • senior present in household

Implementation

Prediction distribution

statistic value
mean 1,500
std dev 682
min 156
25th 967
median 1,399
75th 1,926
max 7,828

External validity

Our model indicates an average monthly housing payment (rent or mortgage) of $1,500 across scored person records. 2024 U.S. Census ACS 1-year estimates (S2503) report a national median monthly gross rent of $1,435.