plotVar: Plot model predictors against the response variable with...

In lfpdroubi/appraiseR: R Package For Real Estate Appraisals

Plot model predictors against the response variable with intervals

Description

plotvar plots model variables with confidence/prediction intervals

Usage

plotVar(
  object,
  variable,
  FUN,
  interval = c("none", "confidence", "prediction", "both"),
  level = 0.8,
  ca = FALSE,
  av,
  at,
  ...
)

Arguments

object	object of class lm
variable	variable to be plotted against response variable
FUN	function used to transform the response (optional)
interval	the type of interval calculation (provided to predict.lm) to be ploted.
level	Tolerance/confidence level (provided to predict.lm) to be ploted.
ca	(T/F) should the limits of the invertal of arbitration be plotted?
av	(T/F) should the arbitrated value be plotted?
at	list to be used for calculate the estimates (defaults for center of each variable).
...	further arguments passed to predict.lm.

Examples

library(sf)
data(centro_2015)
centro_2015 <- within(centro_2015, VU <- valor/area_total)
fit <- lm(log(VU) ~ log(area_total) + quartos + suites + garagens +
            log(dist_b_mar) + padrao,
          data = centro_2015)
plotVar(fit, "area_total")
plotVar(fit, "area_total", residuals = TRUE)
plotVar(fit, "area_total", residuals = TRUE, colour = padrao)
plotVar(fit, "area_total", interval = "confidence", residuals = TRUE)
plotVar(fit, "area_total", interval = "prediction", residuals = TRUE)
plotVar(fit, "area_total", interval = "both", residuals = TRUE)
plotVar(fit, "area_total", interval = "both", ca = TRUE, residuals = TRUE)
plotVar(fit, "area_total", FUN = "log")
plotVar(fit, "area_total", FUN = "log",
        interval = "both", ca = TRUE, residuals = TRUE)
plotVar(fit, "suites")
plotVar(fit, "suites", interval = "confidence")
plotVar(fit, "suites", interval = "confidence", FUN = "log")

# Plot model passing through point 'at'

plotVar(fit, "area_total", interval = "both", ca = TRUE,
        at = list(area_total = 205, quartos = 3, suites = 1,
                  garagens = 2, dist_b_mar = 250, padrao = "medio"))
# Same above + Point valuation at R$ 5.650,00 /m2
plotVar(fit, "area_total", interval = "both",
        at = data.frame(area_total = 205, quartos = 3, suites = 1,
                        garagens = 2, dist_b_mar = 250, padrao = "medio"),
        ca = TRUE, av = log(5650))
# Same above, in the original scale
plotVar(fit, "area_total", interval = "both", FUN = 'log',
        at = data.frame(area_total = 205, quartos = 3, suites = 1,
                        garagens = 2, dist_b_mar = 250, padrao = "medio"),
        ca = TRUE, av = 5650)

# Plotting factors

plotVar(fit, "padrao")
plotVar(fit, "padrao", ca = TRUE)
plotVar(fit, "padrao", FUN = "log", ca = TRUE)
plotVar(fit, "padrao", interval = "confidence")
plotVar(fit, "padrao", interval = "prediction")
plotVar(fit, "padrao", interval = "both")
plotVar(fit, "padrao", FUN = "log", interval = "confidence",
        at = list(area_total = 205, quartos = 3, suites = 1, garagens = 2,
                  dist_b_mar = 250, padrao = "medio"))
plotVar(fit, "padrao", FUN = "log", interval = "prediction",
        at = list(area_total = 205, quartos = 3, suites = 1, garagens = 2,
                  dist_b_mar = 250, padrao = "medio"),
        av = 5650)
# Remove outliers
fit2 <- update(fit, .~.-suites, subset = -c(31,39, 45))
plotVar(fit2, "padrao", FUN = "log", interval = "prediction",
        at = list(area_total = 205, quartos = 3, suites = 1, garagens = 2,
                  dist_b_mar = 250, padrao = "medio"),
        av = 5650)
plotVar(fit2, "area_total", FUN = "log", interval = "both",
        at = list(area_total = 205, quartos = 3, suites = 1, garagens = 2,
                  dist_b_mar = 250, padrao = "medio"),
        ca = TRUE, av = 5650)
plotVar(fit2, "dist_b_mar", FUN = "log", interval = "both",
        at = list(area_total = 205, quartos = 3, suites = 1, garagens = 2,
                  dist_b_mar = 250, padrao = "medio"),
        ca = TRUE, av = 5650)

lfpdroubi/appraiseR documentation built on April 14, 2024, 10:27 p.m.