pif.heatmap: Graphical Sensitivity Analysis of Potential Impact Fraction's...

Description Usage Arguments Value Author(s) See Also Examples

View source: R/pif_heatmap.R

Description

Provides a graphical sensitivity analysis for pif by varying the parameters of a bivariate counterfactual function cft. By default it evaluates the counterfactual:

cft(X) = aX + b.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
pif.heatmap(X, thetahat, rr, cft = function(X, a, b) {     a * X + b },
  method = "empirical", weights = rep(1/nrow(as.matrix(X)),
  nrow(as.matrix(X))), Xvar = var(X), deriv.method.args = list(),
  deriv.method = "Richardson", adjust = 1, n = 512, ktype = "gaussian",
  bw = "SJ", legendtitle = "PIF", mina = 0, maxa = 1, minb = -1,
  maxb = 0, nmesh = 10,
  title = paste0("Potential Impact Fraction (PIF) with counterfactual",
  "\nf(X)= aX+b"), xlab = "a", ylab = "b",
  colors = rev(heat.colors(nmesh)), check_exposure = TRUE,
  check_rr = TRUE, check_integrals = TRUE)

Arguments

X

Random sample (data.frame) which includes exposure and covariates or sample mean if "approximate" method is selected.

thetahat

Asymptotically consistent or Fisher consistent estimator (vector) of theta for the Relative Risk function.

rr

function for Relative Risk which uses parameter theta. The order of the parameters shound be rr(X, theta).

**Optional**

cft

function cft(X, a, b) for counterfactual dependent on one dimensional parameters a and b. Default counterfactual is affine: aX + b.

method

Either "empirical" (default), "kernel" or "approximate". For details on estimation methods see pif.

weights

Normalized survey weights for the sample X.

Xvar

Variance of exposure levels (for "approximate" method).

deriv.method.args

method.args for hessian (for "approximate" method).

deriv.method

method for hessian. Don't change this unless you know what you are doing (for "approximate" method).

adjust

Adjust bandwith parameter (for "kernel" method) from density.

n

Number of equally spaced points at which the density (for "kernel" method) is to be estimated (see density).

ktype

kernel type: "gaussian", "epanechnikov", "rectangular", "triangular", "biweight", "cosine", "optcosine" (for "kernel" method). Additional information on kernels in density.

bw

Smoothing bandwith parameter (for "kernel" method) from density. Default "SJ".

legendtitle

string title for the legend of plot.

mina

Minimum for parameter a for the counterfactual (default 0).

maxa

Maximum for parameter a for the counterfactual (default 1).

minb

Minimum for parameter b for the counterfactual (default -1).

maxb

Maximum for parameter b for the counterfactual (default 0).

nmesh

Number of tiles in plot (default 10).

title

string title for the plot.

xlab

string label for the X-axis of the plot (corresponding to "a").

ylab

string label for the Y-axis of the plot (corresponding to "b").

colors

vector of colours for the heatmap.

check_exposure

boolean Check that exposure X is positive and numeric.

check_rr

boolean Check that Relative Risk function rr equals 1 when evaluated at 0.

check_integrals

boolean Check that counterfactual cft and relative risk's rr expected values are well defined for this scenario.

Value

plotpif ggplot object plotting a heatmap with sensitivity analysis of the counterfactual.

Author(s)

Rodrigo Zepeda-Tello [email protected]

Dalia Camacho-Garc<c3><ad>a-Forment<c3><ad> [email protected]

See Also

See pif for Potential Impact Fraction estimation, pif.sensitivity for sensitivity analysis of the convergence process, pif.plot for a plot of Potential Impact Fraction as a function of the relative risk's parameter theta.

Examples

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
## Not run: 
#Example 1
#------------------------------------------------------------------
X  <- data.frame(rnorm(25,3))            #Sample
rr <- function(X,theta){exp(X*theta)}    #Relative risk
theta <- 0.01                            #Estimate of theta
pif.heatmap(X, theta = theta, rr = rr)

#Save file using ggplot2
#require(ggplot2)
#ggsave("My Potential Impact Fraction Heatmap Analysis.pdf")

#Change pif estimation method to kernel

pif.heatmap(X, theta = theta, rr = rr, method = "kernel")

#Example 2
#------------------------------------------------------------------
X     <- data.frame(Exposure = rbeta(100, 1, 0.2))
theta <- c(0.12, 1)
rr    <- function(X,theta){X*theta[1] + theta[2]}
cft   <- function(X, a, b){sin(a*X)*b}
pif.heatmap(X, theta = theta, rr = rr, cft = cft, 
     nmesh = 15, colors = rainbow(30), method = "empirical",
     title = "PIF with counterfactual cft(X) = sin(a*X)*b")

#Change estimation method to approximate
Xmean <- data.frame(mean(X[,"Exposure"]))
Xvar  <- var(X)
pif.heatmap(Xmean, Xvar = Xvar, theta = theta, rr = rr, cft = cft, 
     nmesh = 15, colors = rainbow(30), method = "approximate",
     title = "PIF with counterfactual cft(X) = sin(a*X)*b")


#Example 3: Plot univariate counterfactuals
#------------------------------------------------------------------
X       <- data.frame(rgamma(100, 1, 0.2))
theta   <- c(0.12, 1)
rr      <- function(X,theta){X*theta[1] + theta[2]}
cft     <- function(X, a, b){sqrt(a*X)}   #Leave two variables in it
pifplot <- pif.heatmap(X, theta = theta, rr = rr, 
 cft = cft, mina = 0, maxa = 1, minb = 0, maxb = 0, 
 legendtitle = "Potential Impact Fraction",
 title ="Univariate counterfactual", ylab = "", colors = topo.colors(10))
pifplot

#You can also add additional ggplot objects
#require(ggplot2)
#pifplot + annotate("text", x = 0.25, y = 0.4, label = "10yr scenario") + 
#geom_vline(aes(xintercept = 0.5), linetype = "dashed") +
#geom_segment(aes(x = 0.25, y = 0.38, xend = 0.5, yend = 0.3), 
#arrow = arrow(length = unit(0.25, "cm")))

#Example 4: Plot counterfactual with categorical risks
#------------------------------------------------------------------
set.seed(18427)
X        <- data.frame(Exposure = 
              sample(c("Normal","Overweight","Obese"), 100, 
                      replace = TRUE, prob = c(0.4, 0.1, 0.5)))
thetahat <- c(1, 1.7, 2)

#Categorical relative risk function
rr <- function(X, theta){

   #Create return vector with default risk of 1
   r_risk <- rep(1, nrow(X))
   
   #Assign categorical relative risk
   r_risk[which(X[,"Exposure"] == "Normal")]      <- thetahat[1]
   r_risk[which(X[,"Exposure"] == "Overweight")]  <- thetahat[2]
   r_risk[which(X[,"Exposure"] == "Obese")]       <- thetahat[3]
   
   return(r_risk)
}

#Counterfactual of reducing a certain percent of obesity and overweight cases
#to normality
cft <- function(X, per.over, per.obese){

   #Find the overweight and obese individuals
   which_obese <- which(X[,"Exposure"] == "Obese")
   which_over  <- which(X[,"Exposure"] == "Overweight")
   
   #Reduce per.over % of overweight and per.obese % of obese
   X[sample(which_obese, length(which_obese)*per.obese),
       "Exposure"] <- "Normal"
   X[sample(which_over,  length(which_over)*per.over),
       "Exposure"] <- "Normal"
   
   return(X)
}

pif.heatmap(X, thetahat = thetahat, rr = rr, cft = cft, mina = 0, minb = 0, maxa = 1, 
            maxb = 1, title = "PIF of excess-weight reduction",
            xlab = "% Overweight cases", ylab = "% Obese cases",
            check_exposure = FALSE, check_rr = FALSE)

## End(Not run)

pifpaf documentation built on Sept. 29, 2017, 1:03 a.m.