Skip to content

Latest commit

 

History

History
249 lines (190 loc) · 8.75 KB

README.md

File metadata and controls

249 lines (190 loc) · 8.75 KB
Raw

README-ContDataQC

#> Last Update: 2022-06-15 17:05:50

ContDataQC

Quality control checks on continuous data. Example data is from a HOBO data logger with 30 minute intervals.

Badges

Maintenance lifecycle License: MIT

CodeFactor codecov R-CMD-check

GitHub issues

GitHub release Github all releases

Installation

To install the current version of the code from GitHub use the example below.

if(!require(remotes)){install.packages("remotes")}  #install if needed
remotes::install_github("leppott/ContDataQC")

The vignette (big help file) isn’t created when installing from GitHub with the above command. If you want the vignette download the compressed file from GitHub and install from that file or install with the command below. The “force = TRUE” command is used to ensure the package will install over and existing install of the same version (e.g., the same version without the vignettes).

if(!require(remotes)){install.packages("remotes")}  #install if needed
remotes::install_github("leppott/ContDataQC", force = TRUE, build_vignettes = TRUE)

If dependent libraries do not install you can install them separately. This happens occasionally. Without all of the packages the main package and/or the vignettes will not install properly. Install the separate packages below and then retry installing ContDataQC.

# libraries to be installed
pkg <- c("remotes"        # install helper for non CRAN libraries
        , "installr"       # install helper
        , "digest"         # caused error in R v3.2.3 without it
        , "dataRetrieval"  # loads USGS data into R
        , "knitr"          # create documents in other formats (e.g., PDF or Word)
        , "doBy"           # summary stats
        , "zoo"            # z's ordered observations, use for rolling sd calc
        , "htmltools"      # needed for knitr and doesn't always install properly with Pandoc
        , "rmarkdown"      # needed for knitr and doesn't always install properly with Pandoc
        , "htmltools"      # a dependency that is sometimes missed.
        , "evaluate"       # a dependency that is sometimes missed.
        , "highr"          # a dependency that is sometimes missed.
        , "rmarkdown"      # a dependency that is sometimes missed.
        , "ggplot2"        # plots
        , "installr"       # used to install Pandoc
        , "rsconnect"
        , "rLakeAnalyzer"
        , "shiny"          # shiny example
        , "survival"
        , "shinyFiles"     # more Shiny
        , "shinyThemes"    # more Shiny
        , "XLConnect"      # read Excel files
        , "zip"            # read/save zip files
        )
#
lapply(pkg, function(x) install.packages(x))

Non-CRAN packages have to be installed separately from GitHub using remotes.

if(!require(remotes)){install.packages("remotes")}  #install if needed
# non-CRAN packages
remotes::install_github("jasonelaw/iha", force = TRUE, build_vignettes = TRUE)
remotes::install_github("tsangyp/StreamThermal", force = TRUE, build_vignettes = TRUE)

Additionally Pandoc is required for creating the reports and needs to be installed separately. If you have RStudio installed it comes with Pandoc and you don’t need to install Pandoc separately.

## pandoc
if(!require(installr)){install.packages("installr")}  #install if needed
installr::install.pandoc()

Purpose

Built for a project for USEPA for Regional Monitoring Networks (RMN).

Takes as input continuous data from data loggers and QCs it by checking for gross differences, spikes, rate of change differences, flat line (consecutive same values), and data gaps. The ContDataQC package provides a organized workflow to QC, aggregate, partition, and generate summary stats.

The code was presented at the following workshops. And further developed under contract to USEPA.

  • Oct 2015, SWPBA (Region 4 regional biologist meeting, Myrtle Beach, SC).

  • Mar 2016, AMAAB (Region 3 regional biologist meeting, Cacapon, WV).

  • Apr 2016, NWQMC (National Water Monitoring Council Conference, Tampa, FL).

Functions were developed to help data generators handle data from continuous data sensors (e.g., HOBO data loggers).

From a single function, ContDataQC(), can QC, aggregate, or calculate summary stats on data. ContDataQC Uses the USGS dataRetrieval library to get USGS gage data. Reports are generated in Word (through the use of knitr and Pandoc).

Usage

Every time R is launched the ContDataQC package needs to be loaded.

# load library and dependant libraries
require("ContDataQC")

The default working directory is based on how R was installed but is typically the user’s ‘MyDocuments’ folder. You can change it through the menu bar in R (File - Change dir) or RStudio (Session - Set Working Directory). You can also change it from the command line.

# if specify directory use "/" not "\" (as used in Windows) and leave off final "/" (example below).
#myDir.BASE  <- "C:/Users/Erik.Leppo/Documents/NCEA_DataInfrastructure/Erik"
myDir.BASE <- getwd()
setwd(myDir.BASE)

Planned Updates

  • Spell out “AW”” and other abbreviations (e.g., AirWater). 20170308. On hold.

  • Gaps in data not always evident in the plots. 20170308.

  • Use futile.logger to better log output for user. Issue #29. 20170606.

  • Debug Aggregate operation. 20170919.

  • Create CDFs. Similar to code already used in previous analyses by Lei. 20170919.

  • PeriodStats(), add number and/or percent of observations above given threshold. 20170919.

  • Fix mixed case issue with filenames in “file” versions of QCRaw, Aggregate, and Stats. 20170929.

    • More check data stuff.
    • Update vignette
    • Threshold number or pct on plot
    • Excel file update (see 9/28/2017 email)

  • PeriodStats, standardize range of y-axis for each time period.

Help

Every function has a help file with a working example. There is also a vignette with descriptions and examples of all functions in the ContDataQC library.

# To get help on a function
# library(ContDataQC) # the library must be loaded before accessing help
?ContDataQC

To see all available functions in the package use the command below.

# To get index of help on all functions
# library(ContDataQC) # the library must be loaded before accessing help
help(package="ContDataQC")

The vignette file is located in the “doc” directory of the library in the R install folder. Below is the path to the file on my PC. But it is much easier to use the code below to call the vignette by name. There is also be a link to the vignette at the top of the help index for the package.

“C:\Programs\R\R-3.4.2\library\ContDataQC\doc\ContDataQC_Vignette.html”

vignette("ContDataQC_Vignette", package="ContDataQC")

If the vignette fails to show on your computer. Run the code below to reinstall the package and specify the creation of the vignette. As noted above if dependent packages are not installed the vignette will fail to build. See above for installing packages.

library(remotes)
install_github("leppott/ContDataQC", force=TRUE, build_vignettes=TRUE)

Guide Videos

Guide videos were created for the ContDataQC package and posted on YouTube.
The Powerpoint slides (pptx), R code notebooks (html), and videos are hosted on a companion GitHub site.

https://github.com/leppott/ContDataQC_Guide

YouTube video links below.