estimates

The family of commands contained under the estimates command are all involved with saving and manipulating estimation results. They are used after a command that fits a model (such as regress) Ultimately, estimates has three sets of abilities that mesh well together.

The first is simply to save estimation results to their own file so that you can access them at your leisure. To save a set of estimation results simply type estimates save [filename]. Then to reuse the file the command is estimates use [filename] much like the commands to use and save normal data. Once you have reloaded a series of estimations, you can use them with any postestimation command you like. It is worth noting however, that Stata no longer knows wht the estimation sample was and will refuse to run any commands that involve returning to the original data (since that may not be loaded). For a more complete rundown of estimates save see help estimates save

The second set of abilities is temporarily storing estimation results in memory rather than to disk. This can be useful if you fit a variety of models and want to be able to switch between them quickly.

Note: When you exit Stata, all estimation results which are being stored will vanish, so make sure you save any you want to keep (see above).

To store an estimation result in memory, simply type estimates store [name] To load the stored estimation you have only to type estimates restore [name] To see if the currently active set of estimations are stored, the command is estimates query which will display the name you've stored it under if it is, in fact, stored. Similarly, estimates dir will list all the estimates currently being stored by name. Finally, you can clear all the estimates from memory using estimates clear or just drop a few with estimates drop [name(s)]

Two additional areas of note for dealing with estimation results are estimates table and estimates stats The latter is used to report model-selection statistics (the AIC and BIC). A basic introduction and links to useful resources can be found on Wikipedia. The second information criterion computed is the BIC or Bayesian Information Criterion. A basic introduction and links to useful resources can be found on Wikipedia. These two criteria are not always in agreement about which model is a better fit.The command itself prints a table for either the most recent estimate (using just estimates stats) or for the stored estimations specified (using estimates stats [name(s)]

The estimates table command is most useful for comparing different estimation results. There are a great many options that can be used with this command. For more information, see help estimates table or see [R] estimates table (Stat's official reference guide).

Back to Estimation