This package provides tools for creating and exporting publication-quality tables in GAUSS. The modern pubtable API is designed for coefficient tables, model comparison tables, summary/statistics tables, and custom matrix/data tables.

Legacy tableControl and tableSet files are retained for migration experiments, but new work should use the pubtable API.

The package now includes an early pubtable API for GAUSS-first publication tables. It keeps table construction, formatting, and export separate, and includes convenience adapters for common GAUSS output structures.

new;
library pubtable;

struct olsmtControl ctl;
struct olsmtOut out;

ctl = olsmtControlCreate;
ctl.output = 0;

out = olsmt(ctl, getGAUSSHome() $+ "examples/auto.dat", "mpg ~ weight + length");

tbl = ptTableFrom(out);
tbl = ptSetTitle(tbl, "OLS Regression");

call ptExport(tbl, "ols_table.md");
call ptExport(tbl, "ols_table.tex");
call ptExport(tbl, "ols_table.xlsx");
call ptExport(tbl, "ols_table.rtf");

Supported table sources in this first version:

Coefficient tables show one statistic row per term (standard errors by default). Use ptModelSetStatRows(model, statRows) (or ptSetStatRows(tbl, statRows)) to choose any combination of "se", "tstat", "pvalue", and "ci"; confidence intervals require calling ptModelSetCI(model, ciLower, ciUpper) first.

Significance stars use the default cutoffs 0.10/0.05/0.01 with symbols "+"/"*"/"**". Use ptModelSetStars(model, cutoffs, symbols) (or ptSetStars(tbl, cutoffs, symbols)) for custom thresholds/symbols, and ptModelNoStars(model) (or ptNoStars(tbl)) to disable stars entirely.

ptModelApplyPreset(model, preset) / ptApplyPreset(tbl, preset) apply a bundle of ptFormat settings in one call:

Apply a preset before calling ptModelTable/ptExport so the chosen formatting is used when the table is rendered.

Exporting/rendering a "journal" or "journal_booktabs" (the default) table with no title set prints a non-fatal errorlog warning ("pubtable warning: journal-style table has no title. Use ptSetTitle(tbl, ...) before exporting.") but still completes the export — call ptSetTitle(tbl, ...) first to silence it.

For more control over model comparisons, build a ptCompareOptions struct with ptCompareOptionsCreate() and pass it to ptModelCompareWith(models, opts):

ptModelCompare(models) is shorthand for ptModelCompareWith(models, ptCompareOptionsCreate()).

ptSetColGroups(tbl, colGroups) sets one column-group label per body column (use "" for ungrouped columns). Contiguous columns with the same label are rendered as a single spanning header:

tbl = ptSetColGroups(tbl, "Sample A" $| "Sample A" $| "Sample B" $| "Sample B");

Markdown, CSV, and plain text render a pseudo-span (the label appears in the first column of the run, with blanks for the remaining spanned columns); LaTeX renders \multicolumn{n}{c}{...} with a \cmidrule(lr){...} underneath; HTML renders <th colspan="n">; RTF renders merged header cells via \clmgf/\clmrg.

ptSetColAlign(tbl, colAlign) / ptModelSetColAlign(model, colAlign) previously only affected LaTeX output. The same colAlign string (one l/c/r per column, including the stub column) now also controls:

Leaving colAlign unset keeps the previous defaults: the stub column left-aligned and data columns right-aligned.

ptSetColFormat(tbl, colDigits) re-formats the body cells of one or more columns to a chosen number of decimal digits (one entry per body column, "" to leave a column unchanged):

tbl = ptSetColFormat(tbl, "0" $| "2");

It re-parses each already-rendered numeric cell with strtof and reformats it with ptFormatNumber, so it works on plain numeric tables but not on cells already wrapped with significance stars or statistic parentheses.

ptSetCellStyle(tbl, row, col, style) marks an individual body cell (1-based row/col) as "bold", "italic", "bold italic", or "":

tbl = ptSetCellStyle(tbl, 1, 2, "bold");

Markdown, LaTeX, HTML, and RTF renderers apply the styling (**bold**, \textbf{...}, <strong>...</strong>, \b ... \b0); CSV and plain text renderers ignore cell styling since those formats can't represent it.

ptExportAll(tables, fname) writes an array of ptTable structs (e.g. reshape(tbl, n, 1) with indexed assignment) to a single file, dispatching on extension like ptExport:

ptExportAllFormats(tables, basename, exts) builds on ptExportAll to produce a full report in several formats with one call:

call ptExportAllFormats(tables, "results/report", "md" $| "tex" $| "html" $| "xlsx");

This writes results/report.md, results/report.tex, results/report.html, and results/report.xlsx, each via ptExportAll. It returns 0 if every format exported successfully, otherwise the return code of the first format that failed (the remaining formats are still attempted).

Initial automatic adapters:

Adapters for optional GAUSS packages (cmlmt, maxlikmt, optmt, tsmt) are activated via pubtable.dec, a declaration file that ships with the library. Run pubtableSet() once after installing pubtable and any optional libraries to write the appropriate #define entries to pubtable.dec:

library pubtable;
call pubtableSet();

Each optional adapter file (pubtable_cmlmt.src, pubtable_maxlikmt.src, pubtable_optmt.src, pubtable_tsmt.src) includes pubtable.dec itself, so user programs don't need to #include anything — just load the optional library together with pubtable in a single statement:

library cmlmt, pubtable;   /* PT_USE_CMLMT is defined because pubtableSet() already ran */

Loading them as two separate statements (library cmlmt; library pubtable;) is not equivalent — each library statement unloads every previously loaded library not named in it, so the second statement would unload cmlmt.

For development installations (git clones not installed via the package manager), run ptSetupAt once with the path to the src/ directory:

call ptSetupAt("C:/path/to/gauss_table_creator/src/");

Re-run pubtableSet() after installing or removing optional libraries.

qardl is auto-detected via the QARDL_SDF_INCLUDED guard in qardl.sdf (its own include guard, not a pubtable.dec entry) — just load it together with pubtable the same way: library qardl, pubtable;.

Once the optional library is loaded this way, most of its model adapters are automatically reachable through the standard ptModelFrom(name, out) dispatcher — no separate function name to remember. The exceptions are optmtResults (no covariance matrix, so there's no ptModel form — use ptTableFromOptmt directly), tscsmtOut (two distinct estimators, no single canonical model — use ptFromTscsmt), and qardlOut/qardlECMOut (need an extra tauIdx argument — use ptFromQardl/ptFromQardlECM).

Optional add-on package adapters and their underlying output structs:

ptModelFromCmlmt/ptModelFromMaxlikmt always compute AIC/BIC from the function value (AIC = -2*fval + 2*k, BIC = -2*fval + 2*k*ln(n)) but keep them hidden by default; call ptModelSetAicBic(model, 1) to reveal the two extra GOF rows. This only ever affects the optional trailing AIC/BIC pair those two adapters add — it never hides an "AIC"/"BIC" GOF row that's already a normal part of another adapter's output (e.g. ptModelFromGlm).

Use ptModelSetDataLabel(model, label) to record a dataset description as its own "Data: <label>." note, kept separate from ptModelSetNotes(model, notes) so callers don't have to hand-concatenate the two.

Initial exporters:

True .docx export is not part of the first version because it requires generating zipped Office Open XML. The practical Word path for now is .rtf/.html, with true .docx a candidate for a later exporter phase.

ptRenderLatex (and ptExport(tbl, "*.tex")) support a few additional ptFormat options:

The program files require a working copy of GAUSS. The modern pubtable API is currently tested with GAUSS 26.

GAUSS 20+ The package can be installed and updated directly in GAUSS using the GAUSS package manager once packaged as pubtable.

GAUSS 18+ Older application-wizard installation is retained as a provisional legacy workflow. If packaging this repository manually, use the pubtable manifest and source files:

1. Zip the package as pubtable.zip.
2. Select Tools > Install Application from the main GAUSS menu.
3. Follow the installer prompts, making sure to navigate to the downloaded pubtable.zip.
4. Before using the package, load the pubtable library:

Note: this installation section is provisional while the package is being modernized.

Modern examples are in examples/model_table_ols.e, examples/model_comparison.e, examples/summary_table.e, examples/summary_statistics_dstatmt.e, examples/export_formats.e, examples/preset_styles.e, examples/addon_tsmt.e, examples/addon_cmlmt.e, and examples/addon_maxlikmt.e. Older tableSet*.e examples are retained as legacy references.