In alekrutkowski/autoschema: Simple diagram of relationships between data.frames

Rd2md::Rd2markdown('man/schema.Rd', 'schema.md')
knitr::opts_chunk$set(echo = TRUE)
N <- 0
schema <- function(...) {
  N <<- N + 1
  cat(DiagrammeRsvg::export_svg(autoschema::schema(...))
      , file=paste0('plot',N,'.svg')) 
}
set.seed(123) # to avoid git changes due to runif()

This is a one-function package. Its sole function autoschema::schema shows the linkages between data.frames and/or csv/tsv files based on their column names and column types/classes. See the usage example below.

The schema function uses the column names and column types/classes of data.frames and/or csv/tsv files to produce a schema diagram of relations between these tables. The inter-connected columns are shaded blue ().
By default, "correct" relations (same column names and same types/classes) are drawn as solid black lines ().
If there are evident type/class mismatches despite the same column names, the relation is drawn as a striped red line ().
If there is a slight mismatch that can be overcome by automatic coercion (integer-double, but both columns numeric) the relation line is dashed but black ().

Dependencies

data.table::fread,  DiagrammeR::grViz,  DiagrammeRsvg::export_svg.

Installation

remotes::install_github('alekrutkowski/autoschema')

Example

Prepare some fake data

DF <- function(...)
  data.frame(...,
             stringsAsFactors=FALSE,
             check.names=FALSE) # to allow for spaces in some column names
FirstDataFrame <-
  DF(RowNum = 1:10,
     Id1 = as.factor(letters[1:10]), # factors are actually integers
     Id2 = letters[1:10],
     Share = seq(0.73, 1.00, 0.03), # as fraction [0,1]
     AnotherVarX = runif(10),
     AnotherVarX2 = runif(10),
     MyDate = as.Date(1e4*runif(10),origin='1899-12-30'))
SecondDataFrame <-
  DF(RowNum = 5:14,
     Id2 = letters[5:14],
     Id1 = letters[5:14],
     Share = as.integer(100*seq(0.73, 1.00, 0.03)), # as % [0,100]
     AnotherVarY = runif(10))
SimpleDF_1 <-
  DF(Id1 = letters[3:12],
     `ID #4` = LETTERS[6:15],
     Share = runif(10),
     `Another Var Q` = runif(10))
SimpleDF_2 <-
  DF(Id2 = letters[3:12],
     `ID #3` = LETTERS[6:15],
     `Another Var S` = runif(10))
write.csv(SimpleDF_1, 'SimpleDF_1.csv',
          row.names=FALSE)
write.table(SimpleDF_2, 'SimpleDF_2.tsv', sep='\t',
            row.names=FALSE)
Last_DF <-
  DF(`ID #4` = LETTERS[2:5],
     `Another Var R` = runif(4))

Plot the linkages

library(autoschema)

Plotting only the minimum necessary linkages (edges); some linkages can be inferred by the viewer indirectly as passing through "intermediary" data.frames:

schema(
  data_frames = list(SecondDataFrame, FirstDataFrame), # or even more data.frames
  data_frame_names = c('Last_DF'), # can be more than 1 data.frame name
  csv_files = c('SimpleDF_1.csv', 'SimpleDF_2.tsv'), # can be more than 2 files
  all_links = FALSE # default
)

Plotting all possible linkages:

schema(
  data_frames = list(FirstDataFrame, SecondDataFrame), # or even more data.frames
  data_frame_names = c('Last_DF'), # can be more than 1 data.frame name
  csv_files = c('SimpleDF_1.csv', 'SimpleDF_2.tsv'), # can be more than 2 files
  all_links = TRUE
)

Warning

The order of the columns within each data.frame is not reflected/preserved in the graphs!

Alternative output options

autoschema::schema can alternatively produce GraphViz DOT code (with output_type='gv') or SVG XML code (with output_type='svg') for further tweaks in the appropriate apps (e.g. https://graphviz.org/resources and https://inkscape.org/develop/about-svg).

alekrutkowski/autoschema documentation built on Dec. 19, 2021, 12:26 a.m.