📝 Table of Contents

• Installation
• Example Code
• Inputs explained for boxcoxar
• R Package Development Reminders
• To-do list

🔨 Installation

Simple

Run

devtools::install_github("daeyounglim/TSEnergy")

Complex

Click the "Clone or download" button and download ZIP. Run the following commands:

unzip("TSEnergy-master.zip")
file.rename("TSEnergy-master", "TSEnergy")
shell("R CMD build TSEnergy")
install.packages(list.files(pattern="(TSEnergy)(.*)(.tar.gz)"), repos = NULL, type = "source")

🔧 Troubleshooting for potential errors

The package runs C++ behind the scenes, and therefore the computer on which this package will be used must have C++ compilers installed beforehand. Otherwise, the installation might fail.

• For Windows users - Download and install Rtools from the website.
• For Mac users - Please install the latest Xcode. It comes with all related compilers.
• For Linux - Install r-base-dev through the package manager of your system. For example, on Ubuntu 18.04, run apt install r-base-dev.

After the compilers are ready, try installing the package again. The dependency installation is supposed to be automated but if R complains, then run the following:

install.packages("Rcpp", "RcppArmadillo", "RcppProgress", "BH")

When the packages above are successfully installed, run devtools::install_github('daeyounglim/TSEnergy') or following the complex version.

Linux distros

When using the devtools::install_github, some Linux distros fire the following error message:

tar: This does not look like a tar archive
gzip: stdin: unexpected end of file

In that case, check if getOption("download.file.method") is curl. Then, run options("download.file.method" = "libcurl") and then try installing again through devtools::install_github("daeyounglim/TSEnergy").

💻 Example Code

######################
# Univariate Box-Cox #
######################
x0 <- read_csv("data-0415_Pharmacy_Biology_Building-10415x_CW_CHWTON-2019-05-09.csv")
x <- x0 %>% filter(TimeStamp >= ymd_hms("2018-07-01 00:00:00"),
                   TimeStamp < ymd_hms("2018-08-01 00:00:00"))
y <- x$Reading %>% scale(center=FALSE) %>% drop
n <- length(y)
x1 <- x$TempRead %>% scale %>% drop
x2 <- x$HumidityRead %>% scale %>% drop
x3 <- 1*(hour(x$TimeStamp) %in% c(0:6, 19:23))

xobs <- cbind(1, x1, x2, x3)
zobs <- cbind(1, x1, x2)
uobs <- cbind(1, x1, x2, x3)

fit <- boxcoxar(y, integer(length(y)), mcmc=list(ndiscard=10000, nkeep = 20000), 
			xobs = xobs, zobs = zobs, uobs = uobs, verbose=TRUE)

## Get the estimates
##  (will be made into 'fitted' function soon)
beta.est <- rowMeans(fit$mcmc.draws$beta)
alpha.est <- rowMeans(fit$mcmc.draws$alpha)
eta.est <- rowMeans(fit$mcmc.draws$eta)

🔣 Inputs explained for boxcoxar

• yobs - a vector of measured energy usage to be used as the response variable
• miss - a vector of binary values (0 or 1), each element of which indicates whether the response variable is missing
• a - the shift parameter for the Box-Cox. By default, set to the minimum value minus one.
• xobs - the covariate matrix for the main regression equation
• zobs - the covariate matrix for the Box-Cox transformation parameter lambda
• uobs - the covariate matrix for the time-dependent variances sigma2
• mcmc - list of MCMC-related parameters: number of burn-ins (ndiscard), number of thinning(nskip), and posterior sample size (nkeep)
• prior - list of hyperparameters; when not given, algorithm will run in default setting
• verbose - logical variable for printing progress bar. Default to FALSE.

👨🏻‍💻 R Package Development Reminders

• When creating an R package, don't forget to register the native symbols by running the following:

tools::package_native_routine_registration_skeleton(".", "src/init.c", character_only=FALSE)

• When put datasets in the package, use usethis::use_data_raw() to create data-raw/ folder and after preparing the dataset, say df, run the following:

devtools::use_data(df)

Then, the convention is to document the datasets in R/data.R in the format of

#' Data from Experiment 1
#'
#' This is data from the first experiment ever to try XYZ using Mechanical
#' Turk workers.
#'
#' @format A data frame with NNNN rows and NN variables:
#' \describe{
#'   \item{subject}{Anonymized Mechanical Turk Worker ID}
#'   \item{trial}{Trial number, from 1..NNN}
#'   ...
#' }
"df"

• Create R/global.R and include all global variables that will be used in the documentation. This includes all

✔️ To do

• MCMC for univariate Box-Cox model
• MCMC for multivariate Box-Cox model
• MCMC for multivariate Box-Cox model with skewness accounted for
• MCMC for multivariate Box-Cox model with heavy-tail accounted for
• MCMC for multivariate Box-Cox model with both skewness and heavy-tail accounted for
• DIC and LPML for each model
• R functions such as fitted, print, summary, plot for each class