Skip to contents

Explore Your Data (ExPanD)

Source: R/ExPanD.R
ExPanD.Rd

A shiny based web app that uses ExPanDaR functionality for interactive data exploration. Designed for long-form panel data but works on simple cross-sectional data as well.

Usage

ExPanD(
  df = NULL,
  cs_id = NULL,
  ts_id = NULL,
  df_def = NULL,
  var_def = NULL,
  config_list = NULL,
  title = "ExPanD - Explore your data!",
  abstract = NULL,
  df_name = deparse(substitute(df)),
  long_def = TRUE,
  factor_cutoff = 10L,
  components = c(sample_selection = TRUE, subset_factor = TRUE, grouping = TRUE,
    bar_chart = TRUE, missing_values = TRUE, udvars = TRUE, descriptive_table = TRUE,
    histogram = TRUE, ext_obs = TRUE, by_group_bar_graph = TRUE, by_group_violin_graph =
    TRUE, trend_graph = TRUE, quantile_trend_graph = TRUE, by_group_trend_graph = TRUE,
    corrplot = TRUE, scatter_plot = TRUE, regression = TRUE),
  html_blocks = NULL,
  export_nb_option = FALSE,
  save_settings_option = TRUE,
  store_encrypted = FALSE,
  key_phrase = "What a wonderful key",
  debug = FALSE,
  ...
)

Arguments

df

A data frame or a list of data frames containing the data that you want to explore. If NULL, ExPanD will start up with a file upload dialog.

cs_id

A character vector containing the names of the variables that identify the cross-section in your data. If only cs_id and not ts_id is provided, the data is treated as cross-sectional, and only appropriate displays are included. df_def overrides if provided.

ts_id

A character scalar identifying the name of the variable that identifies the time series in your data. The according variable needs to be coercible to an ordered vector. If you provide a time series indicator that already is an ordered vector, ExPanD will verify that it has the same levels for each data frame and throw an error otherwise. If cs_id and ts_id are not provided either directly of by df_def, the data is treated as cross-sectional,observations are identified by row names and only appropriate displays are included. df_def overrides if provided.

df_def

An optional dataframe (or a list of dataframes) containing variable names, definitions and types. If NULL (the default) ExPanD uses cs_id and ts_id to identify the data structure and determines the variable types (factor, numeric, logical) based on the classes of the data. See the details section for further information.

var_def

If you specify here a dataframe containing variable names and variable definitions, ExPanD will use these on the provided sample(s) to create the analysis sample. See the details section for the structure of the var_def dataframe. If NULL (default) the sample(s) provided by df will be used as analysis sample(s) directly.

config_list

a list containing the startup configuration for ExPanD to display. Take a look at data(ExPanD_config_russell_3000) for the format. The easiest way to generate a config list is to customize the display within the app and then save the configuration locally.

title

the title to display in the shiny web app.

abstract

An introductory text to display in the shiny web app. Needs to be formatted as clean HTML.

df_name

A character string or a vector of character strings characterizing the dataframe(s) provided in df (will be used in the selection menu of the app)

long_def

If you set this to TRUE (default) and are providing a var_def then ExPanD will add the definitions of the used variables of the underlying dataframe to the definitions provided for the analysis sample to make these more informative to the user. If set to FALSE only the variable definitions provided in the var_def sample will be provided to the user.

factor_cutoff

ExPanD treats factors different from numerical variables. Factors are available for sub-sampling data and for certain plots. Each variable classified as such will be treated as a factor. In addition, ExPanD classifies all logical values and all numerical values with less or equal than factor_cutoff unique values as a factor.

components

A named logical vector indicating the components that you want ExPanD to generate and their order. See the function head of ExpanD for the list of available components. By default, all components are reported. You can also exclude selected components from the standard order by setting then to FALSE. In addition, you can include an arbitrary number of html_block components. Each block will render clean HTML code as contained in the html_blocks parameter below. This allows you to customize your ExPanD report.

html_blocks

A character vector containing the clean HTML code for each html_block that is included in components.

export_nb_option

Do you want to give the user the option to download your data and an R notebook containing code for the analyses that ExPanD displays? Defaults to FALSE.

save_settings_option

Do you want to give the user the option to save and/or load the settings of the ExPanD app to their local environment? Defaults to TRUE.

store_encrypted

Do you want the user-side saved config files to be encrypted? A security measure to avoid that users can inject arbitrary code in the config list. Probably a good idea when you are hosting sensitive data on a publicly available server.

key_phrase

The key phrase to use for encryption. Change this from the default if you want to encrypt the config files.

debug

Do you want ExPanD to echo some debug timing information to the console/log file and to store some diagnostics to the global environment? Probably not.

...

Additional parameters that are passed on to runApp.

Details

If you start ExPanD without any options, it will start with an upload dialog so that the user (e.g., you) can upload a data file for analysis. Supported formats are as provided by the rio package.

When you start ExPanD with a dataframe as the only parameter, it will assume the data to be cross-sectional and will use its row names as the cross-sectional identifier.

When you have panel data in long format, set the ts_id and cs_id parameters to identify the variables that determine the time series and cross-sectional dimensions.

If you provide variable definitions in df_def and/or var_def, ExPanD displays these as tooltips in the descriptive table of the ExPanD app. In this case, you need to identify the panel dimensions in the variable definitions (see below).

When you provide more than one data frame in df, make sure that all have the same variables and variable types defined. If not, ExPanD will throw an error. When you provide only one df_def for multiple data frames, df_def will be recycled.

When you provide var_def, ExPanD starts up in the "advanced mode". The advanced mode uses (a) base sample(s) (the one(s) you provide via df) and the variable definitions in var_def to generate an analysis sample based on the active base sample. In the advanced mode, the app user can delete variables from the analysis sample within the app.

A df_def or var_def dataframe can contain the following variables

"var_name"

Required: The names of the variables that are provided by the base sample or are to be calculated for the analysis sample

"var_def"

Required: For a var_def data frame, the code that is passed to the data frame (grouped by cross-sectional units) in calls to mutate as right hand side to calculate the respective variable. For a data_def data frame, a string describing the nature of the variable.

"type"

Required: One of the strings "cs_id", "ts_id", "factor", "logical" or "numeric", indicating the type of the variable. Please note that at least one variable has to be assigned as a cross-sectional identifier ("cs_id") and exactly one variable that is coercible into an ordered factor has to be assigned as the time-series identifier ("ts_id").

"can_be_na"

Optional: If included, then all variables with this value set to FALSE are required to be non missing in the data set. This reduces the number of observations. If missing, it defaults to being TRUE for all variables other than cs_id and ts_id.

Examples

if (FALSE) {
  ExPanD()

  # Use this if you want to read very large files via the file dialog
  options(shiny.maxRequestSize = 1024^3)
  ExPanD()

  # Explore cross-sectional data
  ExPanD(mtcars)

  # Include the option to download notebook code and data
  ExPanD(mtcars, export_nb_option = TRUE)

  # Use ExPanD on long-form panel data
  data(russell_3000)
  ExPanD(russell_3000, c("coid", "coname"), "period")
  ExPanD(russell_3000, df_def = russell_3000_data_def)
  ExPanD(russell_3000, df_def = russell_3000_data_def,
    components = c(ext_obs = T, descriptive_table = T, regression = T))
  ExPanD(russell_3000, df_def = russell_3000_data_def,
    components = c(missing_values = F, by_group_violin_graph = F))
  ExPanD(russell_3000, df_def = russell_3000_data_def,
    components = c(html_block = T, descriptive_table = T,
    html_block = T, regression = T),
    html_blocks = c(
    paste('<div class="col-sm-2"><h3>HTML Block 1</h3></div>',
    '<div class="col-sm-10">',
    "<p></p>This is a condensed variant of ExPanD with two additional HTML Blocks.",
    "</div>"),
    paste('<div class="col-sm-2"><h3>HTML Block 2</h3></div>',
    '<div class="col-sm-10">',
    "It contains only the descriptive table and the regression component.",
    "</div>")))
  data(ExPanD_config_russell_3000)
  ExPanD(df = russell_3000, df_def = russell_3000_data_def,
    config_list = ExPanD_config_russell_3000)
  exploratory_sample <- sample(nrow(russell_3000), round(0.5*nrow(russell_3000)))
  test_sample <- setdiff(1:nrow(russell_3000), exploratory_sample)
  ExPanD(df = list(russell_3000[exploratory_sample, ], russell_3000[test_sample, ]),
    df_def = russell_3000_data_def,
    df_name = c("Exploratory sample", "Test sample"))
  ExPanD(worldbank, df_def = worldbank_data_def, var_def = worldbank_var_def,
    config_list = ExPanD_config_worldbank)

    }