Type:	Package
Title:	Modern Graphs for Design of Experiments with 'ggplot2'
Version:	0.8
Maintainer:	Jose Toledo Luna <toledo60@protonmail.com>
Description:	Generate commonly used plots in the field of design of experiments using 'ggplot2'. 'ggDoE' currently supports the following plots: alias matrix, box cox transformation, boxplots, lambda plot, regression diagnostic plots, half normal plots, main and interaction effect plots for factorial designs, contour plots for response surface methodology, Pareto plot, and two dimensional projections of a latin hypercube design.
License:	MIT + file LICENSE
Encoding:	UTF-8
LazyData:	true
URL:	https://ggdoe.netlify.app
BugReports:	https://github.com/toledo60/ggDoE/issues
Imports:	ggplot2 (≥ 3.4.0), insight (≥ 0.19.5)
Suggests:	roxygen2, tibble, MASS, viridisLite, ggrepel, patchwork, unrepx, geomtextpath, rsm, lhs, DoE.base
RoxygenNote:	7.3.1
Depends:	R (≥ 3.7.0)
Language:	en-US
NeedsCompilation:	no
Packaged:	2024-02-10 01:28:29 UTC; jtoledo
Author:	Jose Toledo Luna [aut, cre]
Repository:	CRAN
Date/Publication:	2024-02-10 04:50:02 UTC

Adapted epitaxial layer experiment

Description

Same factors and levels as original epitaxial layer experiment but different data

Usage

adapted_epitaxial

Format

A tibble with 16 rows, 4 factors (A,B,C,D), and three responses (ybar,s2,lns2)

Source

Wu, CF Jeff, and Michael S. Hamada. Experiments: planning, analysis, and optimization. John Wiley & Sons, 2011

Color Map on Correlations

Description

Color Map on Correlations

Usage

alias_matrix(
  design,
  midpoint = 0.5,
  digits = 3,
  color_palette = "viridis",
  alpha = 1,
  direction = 1,
  showplot = TRUE
)

Arguments

Value

correlation matrix between main effects and interaction effects from the model.matrix. Alias matrix is also returned

Examples

alias_matrix(design=aliased_design)
alias_matrix(design=aliased_design, color_palette = "plasma")
alias_matrix(design=aliased_design, color_palette = "magma", direction = -1)

D-efficient minimal aliasing design for five factors in 12 runs

Description

D-efficient minimal aliasing design for five factors in 12 runs

Usage

aliased_design

Format

A tibble with 12 rows, 5 factors

Box-Cox Transformations

Description

Box-Cox Transformations

Usage

boxcox_transform(
  model,
  lambda = seq(-2, 2, 1/10),
  showlambda = TRUE,
  lambdaSF = 3,
  showplot = TRUE
)

Arguments

Value

Box-Cox transformation plot with 95% confidence interval of lambda values to consider

Examples

model <- lm(s2 ~ (A+B+C+D),data = adapted_epitaxial)
boxcox_transform(model,lambda = seq(-5,5,0.2))
boxcox_transform(model,lambda = seq(-5,5,0.2),showplot=FALSE)

Convert an object of class 'design' to 'tibble'

Description

Convert an object of class 'design' to 'tibble'

Usage

design_to_tibble(design)

Arguments

Value

Converted design to tibble

Examples

dat <- DoE.base::fac.design(factor.names = list(temp = c(16,32),
time = c(4,12)),replications = 5, randomize = FALSE)
Thk <- c(116.1, 106.7, 116.5, 123.2, 116.9, 107.5, 115.5, 125.1, 112.6, 105.9,
119.2, 124.5, 118.7, 107.1, 114.7, 124, 114.9, 106.5, 118.3, 124.7)
design <- DoE.base::add.response(dat, Thk)
design
design_to_tibble(design)

Boxplots using ggplot2

Description

Boxplots using ggplot2

Usage

gg_boxplots(
  data,
  x,
  y,
  group_var = NULL,
  jitter_points = FALSE,
  horizontal = FALSE,
  point_size = 1,
  alpha = 1,
  color_palette = NA,
  direction = 1,
  show_mean = FALSE
)

Arguments

Value

Boxplots created with ggplot2

Examples

data <- ToothGrowth
data$dose <- factor(data$dose,levels = c(0.5, 1, 2),labels = c("D0.5", "D1", "D2"))
gg_boxplots(data,y= "len",x= "dose",alpha=0.6)
gg_boxplots(data,y = "len",x= "dose",group_var = "supp",
alpha=0.6,color_palette = 'viridis',jitter_points=TRUE)

Regression Diagnostic Plots with ggplot2

Description

Regression Diagnostic Plots with ggplot2

Usage

gg_lm(
  model,
  which_plots = 1:4,
  cooksD_type = 1,
  standard_errors = FALSE,
  point_size = 1.5,
  theme_color = "#21908CFF",
  n_columns = 2
)

Arguments

Details

Plot 5: "Cook's Distance": A data point having a large Cook's distance indicates that the data point strongly influences the fitted values of the model. The default threshold used for detecting or classifying observations as outers is 4/n (i.e cooksD_type=1) where n is the number of observations. The thresholds computed are as follows:

cooksD_type = 1:

4/n

cooksD_type = 2:

4/(n-p-1)

cooksD_type = 3:

1/(n-p-1)

cooksD_type = 4:

3* mean(cook's distance values)

where n is the number of observations and p is the number of predictors.

Plot 6: "Collinearity": Conisders the variance inflation factor (VIF) for multicollinearity:
Tolerance = 1 - R_j^2, VIF = (1/Tolerance) where R_j^2 is the coefficient of determination of a regression of predictor j on all the other predictors. A general rule of thumb is that VIFs exceeding 4 warrant further investigation, while VIFs exceeding 10 indicates a multicollinearity problem

Value

Regression diagnostic plots

References

Belsley, D. A., Kuh, E., and Welsch, R. E. (1980). Regression Diagnostics: Identifying Influential Data and Sources of Collinearity. New York: John Wiley & Sons.

Sheather, S. (2009). A modern approach to regression with R. Springer Science & Business Media.

Examples

model <- lm(mpg ~ wt + am + gear, data = mtcars)
gg_lm(model)

Contour plot(s) of a fitted linear model in ggplot2

Description

Contour plot(s) of a fitted linear model in ggplot2

Usage

gg_rsm(
  rsm_model,
  formula,
  decode = FALSE,
  n_columns = 2,
  text_size = 3,
  bins = 6,
  ...
)

Arguments

Value

A grid of contour plot(s) of a fitted linear model in 'ggplot2'

Examples

## Not run: 
heli.rsm <- rsm::rsm(ave ~ SO(x1, x2, x3),data = rsm::heli)

gg_rsm(heli.rsm,formula = ~x1+x2+x3,at = rsm::xs(heli.rsm),n_columns=3)


## End(Not run)

Girder experiment

Description

An experiment (Narayanan and Adorisio, 1983) to compare four methods for predicting the shear strength for steel plate girders. Data for nine girders in the form of the ratio of predicted to observed load for these procedures are given. Each of the four methods was used to predict the strength of each of the nine girders.

Usage

girder_experiment

Format

A tibble with 36 rows and 3 variables

girders

A factor denoting one of the nine girders

method

A factor denoting one of the four methods for predicting the shear strength for steel plate girders: Aarau, Karlsruhe,Lehigh, Cardiff

response

The shear strength for steel plate girders

Source

Wu, CF Jeff, and Michael S. Hamada. Experiments: planning, analysis, and optimization. John Wiley & Sons, 2011

Examples

lm(response ~ method + girders, data = girder_experiment) |>
anova()

Half-Normal Effects Plots

Description

Half-Normal Effects Plots

Usage

half_normal(
  model,
  method = "Lenth",
  alpha = 0.05,
  label_active = FALSE,
  ref_line = FALSE,
  margin_errors = FALSE,
  point_color = "#21908CFF",
  showplot = TRUE
)

Arguments

Details

The method argument is a simple wrapper for the function PSE() from the unrepx R package. For more details you can use ?unrepx::PSE(). The method arguement implements methods of estimating the standard error of effects estimates from unreplicatd designs. The methods include

Daniel:

The 68.3rd quantile of the absolute effects. See Daniel (1959)

Dong:

The RMS method, applied after excluding all effects that exceed 2.5 * PSE(effects, "SMedian") in absolute value. See Dong (1993)

JuanPena:

An iterated median method whereby we repeatedly calculate the median of the absolute effects that don't exceed 3.5 times the previous median, until it stabilizes. The estimate is the final median, divided by .6578. See Juan and Pena (1992).

Lenth (Default):

The SMedian method, applied after excluding all effects that exceed 2.5 * PSE(effects, "SMedian") in absolute value. See Lenth (1989)

RMS:

Square root of the mean of the squared effects. This is not a good PSE in the presence of active effects, but it is provided for sake of comparisons

SMedian:

1.5 times the median of the absolute effects

Zahn, WZahn:

The Zahn method is the slope of the least-squares line fitted to the first m points of unrepx::hnplot(effects, horiz = FALSE), where m = floor(.683 * length(effects)). (This line is fitted through the origin.) The WZahn method is an experimental version of Zahn's method, based on weighted least-squares with weights decreasing linearly from m - .5 to .5, but bounded above by .65m

Value

A tibble with the absolute effects and half-normal quantiles. A ggplot2 version of halfnormal plot for factorial effects is returned

References

Daniel, C (1959) Use of Half-Normal Plots in Interpreting Factorial Two-Level Experiments. Technometrics, 1(4), 311-341

Dong, F (1993) On the Identification of Active Contrasts in Unreplicated Fractional Factorials. Statistica Sinica 3, 209-217

Hamada and Balakrishnan (1998) Analyzing Unreplicated Factorial Experiments: A Review With Some New Proposals. Statistica Sinica 8, 1-41

Juan, J and Pena, D (1992) A Simple Method to Identify Significant Effects in Unreplicated Two-Level Factorial Designs. Communications in Statistics: Theory and Methods 21, 1383-1403

Lenth, R (1989) Quick and Easy Analysis of Unrelicated Factorials Technometrics 31(4), 469-473

Zahn, D (1975) Modifications of and Revised Critical Values for the Half-Normal Plot. Technometrics 17(2), 189-200

Examples

model <- lm(ybar ~ (A+B+C+D)^4,data=adapted_epitaxial)
half_normal(model)
half_normal(model,method='Zahn',alpha=0.1,ref_line=TRUE,
            label_active=TRUE,margin_errors=TRUE)

Two-Factor interaction effects plot for a factorial design

Description

Two-Factor interaction effects plot for a factorial design

Usage

interaction_effects(
  design,
  response,
  exclude_vars = c(),
  linetypes = c("solid", "dashed"),
  colors = c("#4260c9", "#d6443c"),
  n_columns = 2,
  showplot = TRUE
)

Arguments

Value

interaction effects plot between two factors

Examples

interaction_effects(adapted_epitaxial,response = 'ybar',exclude_vars = c('s2','lns2'))

Lambda Plot: Trace of t-statistics

Description

Lambda Plot: Trace of t-statistics

Usage

lambda_plot(
  model,
  lambda = seq(-2, 2, by = 0.1),
  color_palette = "viridis",
  alpha = 1,
  direction = 1,
  showplot = TRUE
)

Arguments

Value

Lambda plot for tracing t-staitics across different values of lambda (in ggplot2)

Examples

mod <- lm(s2 ~ (A+B+C)^2,data=original_epitaxial)
lambda_plot(mod)
lambda_plot(mod,lambda = seq(0,2,0.1))
lambda_plot(mod,lambda = seq(0,2,0.1),showplot = FALSE)

Obtain main effect plots in a factorial design

Description

Obtain main effect plots in a factorial design

Usage

main_effects(
  design,
  response,
  exclude_vars = c(),
  n_columns = 2,
  color_palette = NA,
  alpha = 1,
  direction = 1,
  showplot = TRUE
)

Arguments

Value

Main effects plots, or a list of tibble with calculated main effects for each factors if showplot=FALSE.

Examples

main_effects(original_epitaxial,response='s2',exclude_vars = c('ybar','lns2'))
main_effects(original_epitaxial,response='ybar',exclude_vars=c('A','s2','lns2'),n_columns=3)

Original epitaxial layer experiment

Description

One of the initial steps in fabricating integrated circuit (IC) devices is to grow an epitaxial layer on polished silicon wafers. The wafers are mounted on a six-faceted cylinder (two wafers per facet), called a susceptor, which is spun inside a metal bell jar. The jar is injected with chemical vapors through nozzles at the top of the jar and heated. The process continues until the epitaxial layer grows to a desired thickness

Usage

original_epitaxial

Format

A tibble with 16 observations, 4 factors (A,B,C,D), and three responses (ybar,s2,lns2)

Factor A

Susceptor-rotation method. Low level is oscillating and high level is continuous

Factor B

Nozzle position. Low level is 2 and high level is 6

Factor C

Deposition temperature (Celsius). Low level is 1210 and high level is 1220

Factor D

Deposition time. Low level is low and high level is high

ybar

average thickness

s2

variance of thickness

lns2

log variance of thickness

Details

In the epitaxial layer growth process, suppose that the four experimental factors, susceptor rotation method, nozzle position, deposition temperature, and deposition time (labeled A, B, Cand D) are to be investigated at the two levels each.

The purpose of this experiment is to find process conditions, that is, combinations of factor levels for A, B, C, and D, under which the average thickness is close to the target 14.5 micrometre with variation as small as possible. The most basic experimental design or plan is the full factorial design, which studies all possible combinations of factors at two levels.

Source

Wu, CF Jeff, and Michael S. Hamada. Experiments: planning, analysis, and optimization. John Wiley & Sons, 2011

Two Dimensional Projections of Latin Hypercube Designs

Description

Two Dimensional Projections of Latin Hypercube Designs

Usage

pair_plots(
  design,
  point_color = "#21908CFF",
  grid = FALSE,
  point_size = 1.5,
  n_columns = 2
)

Arguments

Value

A grid of scatter plots from all two dimensional projections of a Latin hypercube design.

Examples

set.seed(10)
X <- lhs::randomLHS(n=15,k=4)
pair_plots(X,n_columns = 3)

Pareto Plot of Effects

Description

Pareto Plot of Effects

Usage

pareto_plot(
  model,
  alpha = 0.05,
  method = "Lenth",
  effect_colors = c("#F8766D", "#00BFC4"),
  margin_errors = TRUE,
  showplot = TRUE
)

Arguments

Details

The method argument is a simple wrapper for the function PSE() from the unrepx R package. For more details you can use ?unrepx::PSE(). The method arguement implements methods of estimating the standard error of effects estimates from unreplicatd designs. The methods include

Daniel:

The 68.3rd quantile of the absolute effects. See Daniel (1959)

Dong:

The RMS method, applied after excluding all effects that exceed 2.5 * PSE(effects, "SMedian") in absolute value. See Dong (1993)

JuanPena:

An iterated median method whereby we repeatedly calculate the median of the absolute effects that don't exceed 3.5 times the previous median, until it stabilizes. The estimate is the final median, divided by .6578. See Juan and Pena (1992).

Lenth (Default):

The SMedian method, applied after excluding all effects that exceed 2.5 * PSE(effects, "SMedian") in absolute value. See Lenth (1989)

RMS:

Square root of the mean of the squared effects. This is not a good PSE in the presence of active effects, but it is provided for sake of comparisons

SMedian:

1.5 times the median of the absolute effects

Zahn, WZahn:

The Zahn method is the slope of the least-squares line fitted to the first m points of unrepx::hnplot(effects, horiz = FALSE), where m = floor(.683 * length(effects)). (This line is fitted through the origin.) The WZahn method is an experimental version of Zahn's method, based on weighted least-squares with weights decreasing linearly from m - .5 to .5, but bounded above by .65m

Value

A bar plot with ordered effects, margin of error (ME) and simultaneous margin of error (SME) cutoffs.

References

Daniel, C (1959) Use of Half-Normal Plots in Interpreting Factorial Two-Level Experiments. Technometrics, 1(4), 311-341

Dong, F (1993) On the Identification of Active Contrasts in Unreplicated Fractional Factorials. Statistica Sinica 3, 209-217

Hamada and Balakrishnan (1998) Analyzing Unreplicated Factorial Experiments: A Review With Some New Proposals. Statistica Sinica 8, 1-41

Juan, J and Pena, D (1992) A Simple Method to Identify Significant Effects in Unreplicated Two-Level Factorial Designs. Communications in Statistics: Theory and Methods 21, 1383-1403

Lenth, R (1989) Quick and Easy Analysis of Unrelicated Factorials Technometrics 31(4), 469-473

Zahn, D (1975) Modifications of and Revised Critical Values for the Half-Normal Plot. Technometrics 17(2), 189-200

Examples

m1 <- lm(lns2 ~ (A+B+C+D)^4,data=original_epitaxial)
pareto_plot(m1)
pareto_plot(m1,method='Zahn',alpha=0.1)

Reflectance Data, Pulp Experiment

Description

Plant performance is based on pulp brightness as measured by a reflectance meter. Each of the four shift operators (denoted by A, B, C, and D) made five pulp handsheets from unbleached pulp. Reflectance was read for each of the handsheets using a brightness tester

Usage

pulp_experiment

Format

A tibble with 5 rows, and 4 variables (A,B,C,D)

Source

Wu, CF Jeff, and Michael S. Hamada. Experiments: planning, analysis, and optimization. John Wiley & Sons, 2011

Theme for plots used in 'ggDoE'

Description

Theme for plots used in 'ggDoE'

Usage

theme_bw_nogrid()

Value

A simple black and white theme without grid.major and grid.minor for ggplot objects.

Examples

library(ggplot2)
data <- ToothGrowth
data$dose <- factor(data$dose,levels = c(0.5, 1, 2),
                    labels = c("D0.5", "D1", "D2"))

ggplot(data, aes(x=dose, y=len)) +
 geom_boxplot()+
 theme_bw_nogrid()

Simple viridisLite wrapper

Description

Simple viridisLite wrapper

Usage

viridisPalette(
  total_colors,
  color_palette = "viridis",
  alpha = 1,
  direction = 1
)

Arguments

Value

Specified color palette of length 'total_colors'

Examples

viridisPalette(5)
viridisPalette(5,color_palette='magma',alpha=0.5)
viridisPalette(5,color_palette='plasma',alpha=0.6,direction=-1)