Predicting rent with a linear model

Fit monthly rent on floor area and bedrooms for 400 simulated flats, then predict with 95% prediction intervals.

Uses phit

Inputs

Bedrooms Running

fit_lm() fits monthly rent on floor area and bedrooms, as R's lm() would. predict() evaluates the fit on new data, here with 95% prediction intervals for a 2-bedroom flat.

use function Phit\{fit_lm, predict};

$fit = fit_lm('monthly_rent ~ area_m2 + bedrooms', $housing);
$p = predict($fit, ['area_m2' => $areas, 'bedrooms' => $beds], interval: 'prediction');

R²: 0.9333
Residual σ: 154.75
Observations: 400

Coefficients

Term	Estimate	Std. error	t	p	95% CI
(Intercept)	457.0119	23.1998	19.70	< 0.0001	[411.402, 502.622]
area_m2	17.8767	0.4331	41.28	< 0.0001	[17.025, 18.728]
bedrooms	93.4212	12.3749	7.55	< 0.0001	[69.093, 117.750]

n = 400, R² = 0.9333, σ = 154.7470, log-likelihood = -2,582.79

Predicted rent for 2 bedrooms

Predictions

area_m2	fit	lower	upper
30	1,180.15	873.67	1,486.64
45	1,448.3	1,142.94	1,753.67
60	1,716.45	1,411.68	2,021.23
75	1,984.6	1,679.89	2,289.32
90	2,252.75	1,947.56	2,557.95
105	2,520.9	2,214.69	2,827.11
120	2,789.05	2,481.3	3,096.8

Script: site/vignettes/rent-model.php in phpstats/phit@76b22f0.