john-d-fox/gspline
Generalized Regression Splines

Package index
Search the john-d-fox/gspline package
Vignettes
Functions
116
Source code
17
Man pages
21
Home
/
GitHub
/
john-d-fox/gspline
/
panel.fit: Panel functions for predicted values and SE bands

panel.fit: Panel functions for predicted values and SE bands
In john-d-fox/gspline: Generalized Regression Splines

Description Usage Arguments Details Value Functions Author(s) Examples

View source: R/panel-fit.R View source: R/new_gpanel.R

Description

Panel functions for predicted values and SE bands using 'layer' and 'glayer' in the package latticeExtra

This is an experiment in writing a function that can be called via layer or glayer without further complications e.g. xyplot(......,labels = rownames(data)) + layer(panel.labels(...)) or xyplot(....., labels = rownames(data), subscripts = T) + glayer(panel.labels(...)). For selected labels see the examples with trellis.focus and panel.identify

Fills in values in gaps between observed predictor values to help produce a smooth graph of predicted values versus predictor values.

Panel function to display predicted values and SE bands using xyplot in in lattice and layer or glayer in latticeExtra

Usage

  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
 99
100
panel.fit(
  x,
  y,
  fit,
  lower,
  upper,
  subscripts,
  ...,
  type,
  group.number,
  alpha,
  col,
  col.line,
  col.symbol,
  border = F,
  font,
  fontface
)

gpanel.fit(
  x,
  y,
  fit,
  lower,
  upper,
  subscripts,
  ...,
  type,
  group.number,
  alpha,
  col,
  col.line,
  col.symbol,
  border = F,
  font,
  fontface
)

panel.band(
  x,
  y,
  fit,
  lower,
  upper,
  subscripts,
  ...,
  type,
  group.number,
  alpha,
  col,
  col.line,
  col.symbol,
  border = F,
  font,
  fontface
)

panel.errorbars(
  x,
  y,
  fit,
  lower,
  upper,
  length = 0.2,
  angle = 90,
  subscripts,
  ...,
  type,
  group.number,
  alpha,
  col,
  col.line,
  col.symbol,
  border = F,
  font,
  fontface
)

panel.labels(x, y, labels, subscripts, ...)

fillin(data, form, n = 200, xpd = 1)

panel.fit(
  x,
  y,
  fit,
  lower,
  upper,
  subscripts,
  ...,
  type,
  group.number,
  alpha,
  col,
  col.line,
  col.symbol,
  border = F,
  font,
  fontface
)

Arguments

x, y

these are place holders for the arguments that are automatically passed to panel functions from xyplot through layer or glayer. Their values are not used by panel.fit

y

normally passed by 'layer' or 'glayer'. Used only in 'panel.labels'.

fit

fitted values defined as an argument named fit in the call to xyplot and usually passed to panel.fit through layer or glayer

lower, upper

lower and upper limits of error bands, passed in the same way as fit

subscripts

logical, passed xyplot

...

specify any additional arguments to prevent them from being silently passed from xyplot to panel.fit

type

currently not used.

group.number

passed silently from xyplot

alpha

transparency of the bands, in the interval [0,1]. Provide a value or it will be passed silently from xyplot

col

color, may be specified, otherwise passed from xyplot

col.line

line color, if group.number isn't specified.

col.symbol

color when using groups,may be specified, otherwise passed from xyplot

border

if TRUE draw borders on bands. Default is FALSE

font

may be specified, otherwise passed from xyplot

fontface

may be specified, otherwise passed from xyplot

length

(default .2) of crossbar in 'panel.errorbars'

angle

(default 90) of crossbar in 'panel.errorbars'

labels

to display

data

data frame with values of x that need filling in

form

formula idenfying variable x to fill in and grouping variables, g1, g2, etc. using syntax: ~ x + g1 + g2 (the variable to fill in comes first)

n

number of additional points over range of predictor (default 200)

xpd

expansion beyond range of predictor (default 1.0, i.e. no expansion)

X

argument, passed by 'layer' or 'glayer'

dots

passed from main plotting function

Details

With 'layer' and 'glayer' in 'latticeExtra', these functions can be used to easily generate fitted values and confidence or prediction bands that have a reasonable appearance whether a plot uses 'groups' or not.

With layer and glayer in latticeExtra, panel.fit can be used to generate fitted values and confidence or prediction bands that render correctly whether a plot uses groups or not.

If a data frame, data, is used to fit a model and contains variables for fitted values and standard error for estimation or prediction, a typical use of panel.fit has the following form: library(latticeExtra) xyplot( y ~ x, data, groups = g, fit = data$yhat, lower = with(data, yhat - 2*se), upper = with(data, yhat + 2*se), subscripts = T) + glayer(panel.fit(...))

Value

The 'panel.bands', 'panel.fit', 'panel.errorbars', and 'panel.labels' functions are invoked for their graphical effect.

The panel.fit,is invoked for its graphical effect.

Functions

Author(s)

Georges Monette <georges@yorku.ca>

Georges Monette <georges@yorku.ca>

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
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
## Not run: 
  library(gspline)
  library(latticeExtra)
  library(car)
  
  ### Show data, predicted values and confidence bands 
  ### using a 'predict' method that produces 'se'   
  ### -- showing predicted values only at points in the data

  fit <- lm(prestige ~ (income + I(income^2)) * type, Prestige,
      na.action = na.exclude)
  pred <- cbind(Prestige, predict(fit, newdata = Prestige, se = TRUE))
  head(pred)
  (p <- xyplot( prestige ~ income , pred,  groups = type,
                subscripts = T,
                fit = pred$fit,
                lower = with(pred, fit - 2*se.fit),
                upper = with(pred, fit + 2*se.fit)))
  p + glayer(panel.fit(...))

  ###
  ### Use 'fillin' to add points in sparse regions of the predictor
  ### to produce smoother bands
  ###

  fit <- lm( income ~
        (education+I(education^2)+I(education^3) +I(education^4))* type,
        Prestige, na.action = na.exclude)    # overfitting!
  # adding extra values of predictor to get smooth line
  Prestige$occupation <- rownames(Prestige)
  z <- fillin(Prestige, ~education + type, xpd = 1.1) # fill in 'education' within 'type'

  dim(z)
  dim(Prestige)
  z <- cbind(z, predict(fit, newdata = z, se = TRUE))
  head(z)
  gd(3, cex = 2, lwd = 2, alpha = .7)
  (p <-  xyplot( income ~ education, z, groups = type,
                 subscripts = T,
                 fit = z$fit,
                 lower = z$fit - z$se,
                 upper = z$fit + z$se,
                 auto.key = list(space='right', lines= T)))
  p + glayer( panel.fit(...))
  p + glayer( panel.fit(...,  alpha = .1))

  ###
  ###  Prediction with 'wald' for fits with predict
  ###  methods that don't produce se's, e.g. lme
  ###
  
  library(nlme)
  fit <- lme(mathach ~ (ses+I(ses^2)) * Sex * Sector, hs, random = ~ 1|school)
  summary(fit)
  pred <- expand.grid( ses = seq(-2,2,.1), Sex = levels(hs$Sex), Sector = levels(hs$Sector))
  head(pred)
  # merge pred with original data
  hs$from <- 'data'
  pred$from <- 'pred'
  dm <- merge(hs, pred, all = T)
  w <- wald(fit, getX(fit, data = dm)) # attaches data to wald.object so it can included in data frame
  # or:
  w <- wald(fit, pred = dm)  
  w <- as.data.frame(w)
  head(w)
  library(latticeExtra)
  gd(pch = 1, alpha = .2)
  (p <- xyplot(mathach ~ ses | Sector, w, groups = Sex,
       auto.key = T, type = 'p',
       fit = w$coef,
       upper = with(w,coef+2*se),
       lower = with(w,coef-2*se),
       subscript = T))
  p + glayer( gpanel.fit(...,alpha=1))

  wald(fit, 'Sex')  # sig. overall effect of Sex
  wald( fit, ':Sex') # but no evidence of interaction with ses
  wald( fit, '\\^2') # nor of curvature

  # but we continue for the sake of illustration

  L <- Lform( fit, list( 0, 1, 2*ses, 0, Sex == 'Male', (Sex == 'Male')*2*ses), hs)
  L
  (ww <- wald ( fit, L ))
  wald.dd <- as.data.frame(ww, se = 2)
  head( wald.dd )

  xyplot(coef ~ ses | Sex, wald.dd, type = 'n',
      ylim = c(-5, 12),
      ylab = 'increase in mathach per unit increase in ses',
      fit = wald.dd$coef,
      upper = wald.dd$U2,
      lower = wald.dd$L2,
      subscripts = T) +
  layer(panel.fit(...))


  ###
  ###  Using panel.fit with no groups
  ###

  (p <-  xyplot( income ~ education| type, z,
                 subscripts = T,
                 fit = z$fit,
                 lower = z$fit - z$se,
                 upper = z$fit + z$se,
                 auto.key = list(space='right', lines= T)))
  p + layer( panel.fit(...))

  gd_(basecol = 'tomato4')  # Use 'gd_' to set parameters without groups
  p + layer( panel.fit(...))
  p + layer( panel.fit(...,  col = 'grey10'))

  ###
  ### With panels and groups
  ###

  z <- Prestige
  z$gender <- with(z, cut( women, c(-1,15,50,101),labels = c("Male","Mixed","Female")))
  tab(z, ~ gender + type)
  z <- fillin( z, education ~ type + gender, xpd = 1.1)
  fit <- lm( income ~ (education+I(education^2)+I(education^3) )* type * gender,
             z, na.action = na.exclude)    # overfitting!
  summary(fit)
  z <- cbind( z, predict(fit, newdata = z, se = TRUE))
  head(z)
  (p <-  xyplot( income ~ education| gender, z, groups = type,
                 subscripts = T,
                 fit = z$fit,
                 lower = z$fit - z$se,
                 upper = z$fit + z$se,
                 layout = c(1,3),
                 auto.key = list(space='right', lines= T, cex = 1.5)))

  p + glayer(panel.fit(...))
  trellis.focus()
  panel.identify(labels= z$occupation)
  trellis.unfocus()
  z$type2 <- with( z, reorder(type,education, mean, na.rm=T))
  gd(3)
  (p <-  xyplot( income ~ education| type2, z, groups = gender,
                 subscripts = T,
                 fit = z$fit,
                 lower = z$fit - z$se,
                 upper = z$fit + z$se,
                 layout = c(1,3),
                 par.strip.text = list(cex = 2),
                 auto.key = list(space='right', lines= T, cex = 1.5)))

  p + glayer( panel.fit(...))
  trellis.focus()
  panel.identify(labels= z$occupation)
  trellis.unfocus()

  ###
  ### With panels^2
  ### need to remove 'col = col.line'
  ###

  z <- Prestige
  z$occ <- rownames(Prestige)
  z$gender <- with(z, cut( women, c(-1,15,50,101),labels = c("Male","Mixed","Female")))
  z$type2 <- with( z, reorder(type,education, mean, na.rm=T))
  tab(z, ~ gender + type2)
  z <- fillin( z, education ~type + gender, xpd = 1.1)
  fit <- lm( income ~ (education+I(education^2)+I(education^3) )* type * gender,
             z, na.action = na.exclude)    # overfitting!
  summary(fit)
  z <- cbind( z, predict(fit, newdata = z, se = TRUE))
  head(z)
  (p <-  xyplot( income ~ education| gender*type, z,
                 subscripts = T,
                 fit = z$fit,
                 labels = z$occ,
                 lower = z$fit - z$se,
                 upper = z$fit + z$se,
                 auto.key = list(space='right', lines= T, cex = 1.5)))

  p + layer( panel.fit(...))
  p + layer( panel.fit(..., col = 'black', alpha = .1)) + layer(panel.labels(...))
  
  
  

## End(Not run)
## Not run: 
trellis.focus()
panel.identify(labels = rownames(data),rot=-15,col = col.symbol, etc.)
trellis.unfocus()

## End(Not run)

## Not run: 

  ###
  ### Exploring possible discontinuity in value of post-secondary education
  ###


## End(Not run)
## Not run: 
trellis.focus()
panel.identify(labels = rownames(data),rot=-15,col = col.symbol, etc.)
trellis.unfocus()

## End(Not run)

john-d-fox/gspline documentation built on Sept. 17, 2020, 3:19 a.m.

Related to panel.fit in john-d-fox/gspline...

john-d-fox/gspline index
README.md

R Package Documentation

rdrr.io home R language documentation Run R code online

Browse R Packages

CRAN packages Bioconductor packages R-Forge packages GitHub packages

We want your feedback!

Note that we can't provide technical support on individual packages. You should contact the package authors for that.
GitHub issue tracker
ian@mutexlabs.com
Personal blog

 
Embedding an R snippet on your website

Add the following code to your website.

For more information on customizing the embed code, read Embedding Snippets.

Close