dot-get_plot: Predictive Plot for Model
In christithomp/scamplot: Predictive Plots for Shape-Constrained Additive Models

Description Usage Arguments Value Examples

Predictive Plot for Model

1	.get_plot(x, fHat, lb, ub, title = "Prediction Plot", x_lab, y_lab, pred_type)

`x`	- vector containing spline terms
`fHat`	- vector of prediction values
`lb`	- vector containing lower bound of confidence interval
`ub`	-vector containing upper bound of confidence interval
`title`	- optional string for title of the plot
`x_lab`	- string for title of x axis
`y_lab`	- string for title of y axis
`pred_type`	- string indicating "link" or "response"

None

# Load required packaged
library(mgcv); library(HRW)
# Load the data
data(BostonMortgages)

#Create model for data
fit = gam(deny ~ black + s(dir), family = binomial, data = BostonMortgages)

#Create data frame for prediction
ng = 1001
dir_seq = seq(min(BostonMortgages$dir), max(BostonMortgages$dir), length = ng)
black = BostonMortgages$black
black_seq = rep(unique(black)[which.max(tabulate(match(black, unique(black))))], ng)

#Get predicted y values, lower bound, and upper bound of confidence interval
newdata = data.frame(dir = dir_seq, black =black_seq)
pred_fit = predict(fit, newdata, se.fit = TRUE)
fHat = pred_fit$fit
lb = fHat - qnorm(0.975) * pred_fit$se.fit
ub = fHat + qnorm(0.975) * pred_fit$se.fit

#Plot the predictive plot
.get_plot(dir_seq, fHat, lb, ub, x_lab = "dir", y_lab = "deny", pred_type = "link")
.get_plot(dir_seq, fHat, lb, ub, x_lab = "dir", y_lab = "deny", pred_type = "response")