Command Reference Index¶

TabDat-Explore includes 69 commands and topics organized into functional categories. Click any command name to view its syntax, options, and examples.

Load & Inspect¶

Command	Purpose
`use`	Load a dataset from Parquet, Stata `.dta`, CSV, Arrow, or activate a named table.
`describe`	Show dataset shape, variable names, and DuckDB/Arrow data types.
`summarize`	Descriptive statistics (count, mean, std dev, min, max) for numeric variables.
`codebook`	Detailed variable profiling with missingness and unique sample values.
`count`	Count rows in the active dataset.
`head`	Preview the first rows of the active dataset.
`tail`	Preview the last rows of the active dataset.
`status`	Show execution backend state, materialization status, and active relation details.
`doctor`	Inspect environment, core engines, and capability health.

Transform & Subset¶

Command	Purpose
`keep`	Keep specific variables or rows matching a boolean condition.
`drop`	Drop specific variables or rows matching a boolean condition.
`select`	Select a subset of columns into the active dataset.
`generate`	Create a new variable from an arithmetic or logical expression.
`replace`	Replace values in an existing variable conditionally or unconditionally.
`rename`	Rename variables in the active dataset.
`recode`	Recode values or ranges into new categories.

Combine & Reshape¶

Command	Purpose
`join`	Join the active dataset with a named table or file on key variables.
`append`	Vertically stack rows from a named table to the active dataset.
`reshape`	Reshape data between wide and long layouts.

Summarize & Tabulate¶

Command	Purpose
`tabulate`	One-way and two-way frequency tables and crosstabs.
`collapse`	Grouped aggregations (mean, sum, min, max, count) into a new dataset.
`by`	Execute a command independently within groups of variables.

Linear & Quantile Models¶

Command	Purpose
`regress`	Linear regression (OLS, WLS, GLS) with robust and clustered covariance.
`qreg`	Quantile regression for median and conditional quantiles.

Binary & Limited Dependent Variables¶

Command	Purpose
`logit`	Logistic regression via maximum likelihood.
`probit`	Probit regression via maximum likelihood.
`tobit`	Tobit censored regression with lower and upper bounds.
`heckman`	Heckman two-step sample selection estimator.
`nl`	Nonlinear least squares regression.

Count & Survival Models¶

Command	Purpose
`poisson`	Poisson log-linear count regression.
`nbreg`	Negative binomial count regression.
`zip`	Zero-inflated Poisson count regression.
`zinb`	Zero-inflated negative binomial regression.
`streg`	Parametric survival regression (Weibull, Exponential, Cox).

Panel, IV & Causal Inference¶

Command	Purpose
`panel`	Set, display, or clear panel identifier and time metadata.
`ivregress`	Instrumental variables regression (2SLS and GMM).
`cfregress`	Control function regression for endogeneity.
`xtreg`	Panel data fixed-effects (FE) and random-effects (RE) models.
`xtdata`	Panel within and between transformations.
`xtlogit`	Fixed-effects conditional logit regression.
`xtabond`	Arellano-Bond dynamic panel GMM estimator.
`did`	Difference-in-differences estimator with panel metadata.
`drdid`	Doubly robust difference-in-differences estimator.

Machine Learning & Regularization¶

Command	Purpose
`lasso`	L1-regularized Lasso linear regression.
`postlasso`	Post-selection OLS estimation after Lasso.
`ridge`	L2-regularized Ridge linear regression.
`elasticnet`	Elastic net regression with combined L1/L2 penalties.
`cvlasso`	Cross-validated Lasso with optimal penalty selection.
`cvridge`	Cross-validated Ridge regression.
`cvelasticnet`	Cross-validated Elastic net regression.
`dml`	Double / debiased machine learning average treatment effect (ATE).

Bayesian & Spatial¶

Command	Purpose
`bayes`	Bayesian linear regression via Ridge estimation.
`bayes_prefix`	MCMC sampling prefix for linear and logistic regression models.
`bayesplot`	MCMC chain diagnostic plots (trace, density, autocorrelation).
`spregress`	Spatial lag (SAR) and spatial error (SEM) regression models.

Post-Estimation & Hypothesis Tests¶

Command	Purpose
`predict`	Compute fitted values (`xb`), residuals, probabilities (`pr`), or draws.
`estat`	Post-estimation diagnostics (VIF, AIC/BIC, Hausman test, etc.).
`estat_report`	Generate self-contained HTML regression diagnostic report.
`test`	Linear hypothesis Wald test after regression.
`lincom`	Estimate linear combinations of model parameters.
`ttest`	Student t-tests for one sample, two samples, or paired observations.

Visualization¶

Command	Purpose
`histogram`	Save frequency or density histogram plots.
`scatter`	Save bivariate scatter plots with optional fit lines.
`bar`	Save categorical bar charts.

Scripts, SQL & System¶

Command	Purpose
`sql`	Run DuckDB SQL queries against the active dataset.
`run`	Execute a `.td` script file.
`save`	Save the active dataset to Parquet.
`export`	Export the active dataset to Parquet.
`set`	Configure session runtime settings (`graph_format`, `artifact_dir`, `graph_open`).
`lowess`	Locally weighted scatterplot smoothing (LOWESS).
`help`	Display in-app help for commands and topics.
`exit` / `quit`	Exit the interactive shell.