From first look to reproducible report.

DataSum is an R toolkit for rigorous first-pass data diagnostics. It helps statisticians, researchers, professors, scientists, and students move from a raw data frame to transparent summaries, quality warnings, distribution checks, group comparisons, and reproducible reports.

Release status: DataSum 1.0.0 is available from CRAN, the GitHub v1.0.0 release, and the Zenodo archive.

Start in 60 seconds

install.packages("DataSum")

# Development version:
# install.packages("remotes")
# # remotes::install_github("Uzairkhan11w/DataSum")
install.packages("DataSum", repos = c('https://uzairkhan11w.r-universe.dev', 'https://cloud.r-project.org'))

library(DataSum)

summary <- summarize_data(iris, by = "Species", digits = 3)
profile <- profile_data(iris)
profile$warnings

What DataSum gives you

Capability	What it answers
NA-aware summaries	How much usable data is present in every variable?
Robust statistics	What do median, IQR, MAD, skewness, and excess kurtosis reveal?
Mode handling	Are there tied modes, and how frequent are they?
Outlier diagnostics	Which variables exceed the transparent 1.5 x IQR rule?
Normality diagnostics	Which test ran, what was its p-value, and what does the decision mean?
Grouped profiles	How do variables differ across treatments, classes, or cohorts?
Analyst warnings	Which missingness, duplicate, outlier, or distribution issues need attention?
Reproducible reports	Can the diagnostic record be shared as Quarto HTML, PDF, or DOCX?
Browser app	Can someone inspect missingness, summaries, and distributions without installing R?

Clean 1.0 API

Function	Purpose
`summarize_vector()`	One-row diagnostic summary for a single vector
`summarize_data()`	One row per variable, optionally within groups
`profile_data()`	Dataset overview, variable summaries, and warnings
`datasum_report()`	Quarto diagnostic report source and optional rendering
`run_datasum_app()`	Interactive Shiny interface

Try the diagnostics

summarize_vector(
  c(12, 14, 14, 16, NA, 21, 45),
  name = "response_time",
  digits = 2
)

summarize_data(iris, by = "Species", digits = 2)

profile <- profile_data(airquality, digits = 2)
profile$dataset
profile$summary
profile$warnings

Normality output is deliberately cautious. DataSum reports evidence against normality or no evidence against normality; it does not claim that a sample has proven a population distribution.

Browser app

Launch DataSum in your browser →

Explore a CSV without installing R. DataSum calculates summaries, missingness, IQR outlier flags, and charts locally in your browser; your uploaded dataset is not sent to DataSum.

For normality tests, grouped profiles, and reproducible Quarto reports, install the full DataSum R package from CRAN.

Create a reproducible report

Create a portable Quarto source file without extra software:

report <- datasum_report(
  iris,
  path = "iris-diagnostic-report.qmd",
  format = "qmd",
  render = FALSE
)

With the optional quarto package and Quarto CLI installed, render directly:

datasum_report(
  iris,
  path = "iris-diagnostic-report.html",
  format = "html",
  render = TRUE
)

The report contains the dataset overview, variable diagnostics, analyst warnings, formula definitions, and interpretation guidance.

Designed for trust

Tested by GitHub Actions on current R releases for Linux, Windows, and macOS
Deterministic tied-mode output
Safe behavior for missing, empty, constant, and non-numeric vectors
Explicit formulas and thresholds
No silent two-decimal rounding
Source, tests, documentation, and roadmap kept in public

Project links

Citation

GitHub displays a Cite this repository button from CITATION.cff.

Preferred citation for the exact v1.0.0 software archive

Khan, Uzair Javid, Immad Ahmad Shah, and Sukhdev Mishra (2026). DataSum: Modern Data Summaries and Diagnostic Reports for Statistical Analysis (Version 1.0.0) [R package]. Zenodo. https://doi.org/10.5281/zenodo.21352624

Exact version DOI: 10.5281/zenodo.21352624
All-versions Zenodo DOI: 10.5281/zenodo.21352623
CRAN package DOI: 10.32614/CRAN.package.DataSum

Use the version DOI for a reproducible scholarly citation and the concept DOI when referring to DataSum across all releases. From R, you can also run:

citation("DataSum")

Community

DataSum is being built in public. Bug reports, statistical-method discussions, teaching use cases, documentation improvements, and research workflow ideas are welcome through GitHub Issues.

DataSum is diagnostic software, not a substitute for study design, domain expertise, or model-specific assumption checking.