The hardware and bandwidth for this mirror is donated by dogado GmbH, the Webhosting and Full Service-Cloud Provider. Check out our Wordpress Tutorial.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]dogado.de.

agrobox 0.4.0

New features

• New automatic route selection in agrobox(). The goal is to place post-hoc letters through a viable and defensible statistical route. For each cluster the function checks sufficient data, Shapiro-Wilk (residual normality) and Fligner-Killeen (homogeneity of variances), and then selects:

  • Route A (normal + homogeneous): ANOVA + Duncan / Tukey (block and factor2 included in the model).
  • Route B (normal + heteroscedastic): Welch ANOVA + Games-Howell.
  • Route C (not normal + homogeneous): Kruskal-Wallis, or Friedman when a block is present.
  • Route D (not normal + heteroscedastic): same tests as route C, flagged with a note on heterogeneous variances.

• Added non-parametric analysis:

  • Kruskal-Wallis with letters from agricolae::kruskal() or Dunn post-hoc test, selected with the new argument np_test = c("kruskal", "dunn").
  • Friedman test for blocked designs (RCBD). Replicates per treatment x block are averaged and incomplete blocks are excluded. Letters are protected by the global Friedman test (all treatments share “a” when it is not significant).
  • New argument p.adj (default "bonferroni") for p-value adjustment in Kruskal-Wallis and Dunn comparisons.

• Fallback chain: if the post-hoc test of the selected route cannot be computed, the next viable route is tried (A/B -> non-parametric, Friedman -> Kruskal-Wallis, Kruskal-Wallis <-> Dunn). Means without letters are shown only as a last resort, always with an explicit note.

• Each agrobox() figure now includes a method note (caption) describing the route used in each panel: why it was selected, its advantages, limitations and scope. It can be removed with + ggplot2::labs(caption = NULL).

• Route B now computes the Welch ANOVA p-value (oneway.test()).

• New columns in $stats: ruta, metodo, p_ruta (p-value of the global test used by the route) and nota.

• Added agroppt() to export a ggplot figure to a single editable PowerPoint slide (vector graphic), with custom width and height.

Bug fixes

• Games-Howell letters were relabeled as whole strings, so a group such as “ab” became a new letter and produced wrong groupings. Letters are now relabeled letter by letter.
• Games-Howell failed silently when bloque or factor2 were supplied (letters disappeared). It now compares treatments only and adds a note.
• agrobox() failed with could not find function “%>%” when dplyr was not loaded. %>% and .data are now imported.
• estructura = "row~" (documented) produced a parsing error in facet_grid(); it is now completed internally as "row ~ .".
• Group labels containing "_" (or non-ASCII characters in non-UTF-8 sessions) sent means and letters to an extra NA panel. Grouping columns are now recovered by joining on the cluster instead of splitting its label.
• Negative responses: the automatic upper y-axis limit fell below the maximum and removed observations; the CV was reported as negative. The CV now uses the absolute value of the mean.
• Manual lim_sup / lim_inf removed observations outside the range and altered the boxplots. Limits now use coord_cartesian() (zoom only).
• Statistical power was underestimated with blocks or factor2; it now uses partial eta squared (identical results in one-way models).
• Numeric levels (e.g. doses 0, 50, 100, 150) were ordered as text; they are now ordered numerically, and factor levels are respected.
• $tabla columns and $stats rows now follow the facet order defined by grupo1_orden / grupo2_orden.
• Empty cells are shown instead of NA in $tabla when treatments are nested within groups.
• LaTeX \makecell headers in $tabla no longer split labels that contain "_".

Improvements

• A message now lists factor levels that are excluded because they are not included in orden_factor, grupo1_orden or grupo2_orden.
• CV and Power are reported in every route (computed from the classic ANOVA fit).
• The ANOVA row of $tabla shows significance stars only for route A; other routes show "-".
• Example 5 of agrobox() wrapped in \donttest{} to keep example run time within CRAN limits.

Deprecated

• var.equal no longer changes the results: the route is selected automatically. The argument is kept for backward compatibility and a message is shown when it is supplied.

Changes in results compared with 0.3.0

• Panels that previously showed means without letters (e.g. failed normality) now receive letters through routes B, C or D.
• Games-Howell letters may differ where groups overlap (bug fix above).
• Statistical power changes in models with block or factor2.
• Route A results (means, letters, ANOVA, CV) are unchanged.

Dependencies

• Added officer and rvg to Suggests (only required by agroppt()).

agrobox 0.3.0

New features

• Added support for combined ordering and relabeling through:
  • orden_factor
  • grupo1_orden
  • grupo2_orden
  These arguments now accept both named and unnamed vectors:
  • Unnamed vectors reorder levels.
  • Named vectors reorder and relabel simultaneously.
• Added two real-world datasets:
  • nitrogeno_liberacion: nitrogen release dynamics from fertilizers.
  • pimiento_hibridacion: hybrid seed production dynamics in pepper.
• Added new outputs to agrobox():
  • $data: summarized dataset used in the analysis (means, sd, n).
  • $stats: ANOVA diagnostics per cluster (p-values, CV, Power).

Improvements

• Faceting now correctly respects factor level order defined by:

  • grupo1_orden
  • grupo2_orden

  This resolves previous inconsistencies with facet_grid() ordering.

• Improved handling of factor levels across all internal steps:

  • Consistent propagation of levels from raw data → analysis → plotting.
  • Prevents unintended reordering in ggplot facets.

• Cluster construction is now more robust:

  • Uses interaction() when both grouping variables are present.
  • Preserves factor levels in all grouping scenarios.

• Enhanced statistical robustness:

  • Improved handling of missing values (NA) in post-hoc comparisons.
  • Automatic suppression of letters when statistical assumptions fail.

• Improved compatibility with ggplot2 faceting behavior by enforcing factor levels prior to plotting.

Internal changes

• Refactored factor handling through aplicar_orden_labels():
  • Unified logic for ordering and relabeling across all variables.
• Reworked cluster generation logic to support:
  • Single grouping
  • Dual grouping
  • No grouping (single panel)
• Improved reconstruction of facet variables for geom_text():
  • Ensures correct placement of labels within panels.
• Cleaner separation between:
  • data preprocessing
  • statistical analysis
  • visualization
  • reporting

agrobox 0.2.1

Minor fixes

• Resubmission to CRAN after incoming checks.
• Moved magick, kableExtra, and tinytex from Imports to Suggests.

• No user-facing changes.

agrobox 0.2.0

New features

• Added agrosintesis(), a high-level wrapper to automate multi-variable ANOVA summaries using agrobox() as the core engine.
  • Supports multiple response variables.
  • Supports optional clustering through a formula interface (e.g. Variedad ~ Localidad).
  • Automatically iterates over cluster combinations and returns publication-ready summary tables.
  • When estructura = NULL, the analysis is performed globally (no clustering).
• Added agroexcel() to export agrobox and agrosintesis results to Excel (.xlsx) files.
  • Works with single tables and lists of tables.
  • Automatically creates one worksheet per cluster.
  • Sheet names are sanitized to comply with Excel and Windows limitations.
• Added agrotabla() to export result tables as high-resolution PNG images.
  • Uses LaTeX for high-quality table rendering.
  • Supports both single tables and lists of tables.
  • File names are sanitized for LaTeX and filesystem compatibility.
  • Designed for fast reporting and presentation-ready outputs.

Improvements

• Cluster labels now preserve original factor levels instead of numeric codes.
• Improved handling of factor levels when clustering is enabled.
• Column names are automatically escaped to avoid LaTeX compilation issues.
• File and sheet name sanitization improves Windows compatibility.

Internal changes

• Refactored reporting logic to separate analysis from export utilities.
• Avoided side effects in core analytical functions.
• Improved CRAN compliance by:
  • Using fully qualified namespace calls.
  • Avoiding library() calls inside functions.
  • Wrapping file-generating examples in \dontrun{}.

agrobox 0.1.1

Changes

• Improved main function performance and robustness.

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.
Health stats visible at Monitor.