<\/video><\/figure>\n\n\n\nI have no idea why and how, but this turns out to be handy when you need to edit several lines of your code in a similar manner.<\/p>\n\n\n\n
Next let’s talk about two things that I surprisingly had no idea about until this year. I’m grouping them as for me they serve the same purpose: Minimize unnecessary Stata outputs to the hard drive.<\/strong><\/p>\n\n\n\nSeasoned Stata users might know where this is leading to: Temp files. <\/p>\n\n\n\n
Temp files are another kind of Stata macro that stands for placeholder file names. They are treated akin to locals – which means that if you save something in a temp file, Stata stores said stuff temporarily somewhere in its root folder, associates this temporarily stored stuff with the name assigned to the tempfile macro, and deletes the temporary file immediately when the code block is done executing. I find them handy for complicated data-cleaning work<\/strong> where I need to dice up the dataset in different ways and stitch them back together.<\/strong><\/p>\n\n\n\nSo two things: 1. This does not stop the file from ever being stored on your hard drive. But at least it no longer lies there forever! 2. Temp files are local<\/strong>s, so they are “gone” gone after running the code (Otherwise it defies the purpose isn’t it?).<\/p>\n\n\n\nAlright, enough talking. Here is a sketch of how temp files can be used in said setting:<\/p>\n\n\n\n
Stata<\/span><\/span><\/path><\/path><\/svg><\/span>\/\/ Let's assume the working directory has been set and the dataset is just called "data"<\/span><\/span>\nuse<\/span> data, <\/span>clear<\/span><\/span>\ntempfile<\/span> <\/span>temp_data<\/span> <\/span><\/span>\n<\/span>\n\/* Insert whatever data cleaning code here *\/<\/span><\/span>\n<\/span>\nsave<\/span> <\/span>`<\/span>temp_data<\/span>'<\/span><\/span>\n<\/span>\n\/\/ If later down the track you'll need to merge the saved temporary dataset with something else, it can be called just like any other local!<\/span><\/span>\nmerge<\/span> <\/span>1<\/span>:<\/span>1<\/span> id using <\/span>`<\/span>temp_data<\/span>'<\/span><\/span>\nsave<\/span> The_actual_dataset, replace<\/span><\/span><\/code><\/pre><\/div>\n\n\n\nThe handy thing with temp files is that they can be used to store formats other than .dta files as well! Here is our second gimmick which solves one of my long-standing questions: How do you export a single .pdf file with all the graphs you wanted in Stata?<\/p>\n\n\n\n
It turns out that there is a set of commands called putpdf<\/strong> out there. Those commands quite literally put stuff into a (yet to be formally saved) .pdf file. Here I think things will make more sense if I start with a hypothetical example:<\/p>\n\n\n\nStata<\/span><\/span><\/path><\/path><\/svg><\/span>\/\/ Imagine that the dataset has already been loaded and we have a list of variables that each need a) A histogram; b) A kernel density plot.<\/span><\/span>\nlocal<\/span> <\/span>var_list<\/span> <\/span>\/* Insert the variables here *\/<\/span><\/span>\n<\/span>\n******************************************<\/span><\/span>\n* Plotting<\/span><\/span>\n******************************************<\/span><\/span>\nputpdf <\/span>clear<\/span> <\/span>\/\/ Makes sure that there is no existing .pdf file in the background<\/span><\/span>\nputpdf begin <\/span>\/\/ Creates an empty .pdf file<\/span><\/span>\n<\/span>\nputpdf paragraph <\/span>\/\/ Tells Stata "We'll add content from here"<\/span><\/span>\n<\/span>\ntempfile<\/span> <\/span>hist_temp<\/span><\/span>\ntempfile<\/span> <\/span>kern_temp<\/span><\/span>\n<\/span>\nforeach<\/span> <\/span>var<\/span> <\/span>in<\/span> varlist <\/span>`<\/span>var_list<\/span>'<\/span> <\/span>{<\/span><\/span>\n <\/span>hist<\/span> <\/span>`<\/span>var<\/span>'<\/span> <\/span>\/\/ Just vanilla code here as this is not the emphasis<\/span><\/span>\n <\/span>graph<\/span> export <\/span>"`<\/span>hist_temp<\/span>'"<\/span>, <\/span>as<\/span>(<\/span>png<\/span>) replace<\/span><\/span>\n putpdf image <\/span>"`<\/span>hist_temp<\/span>'"<\/span><\/span>\n <\/span><\/span>\n <\/span>kdensity<\/span> <\/span>`<\/span>var<\/span>'<\/span> <\/span>\/\/ Just vanilla code here as this is not the emphasis either<\/span><\/span>\n <\/span>graph<\/span> export <\/span>"`<\/span>kern_temp<\/span>'"<\/span>, <\/span>as<\/span>(<\/span>png<\/span>) replace<\/span><\/span>\n putpdf image <\/span>"`<\/span>kern_temp<\/span>'"<\/span><\/span>\n <\/span><\/span>\n putpdf pagebreak <\/span>\/\/ Adds a pagebreak<\/span><\/span>\n\tputpdf paragraph<\/span><\/span>\n<\/span>\n}<\/span><\/span>\n<\/span>\nputpdf <\/span>save<\/span> <\/span>"Plots.pdf"<\/span>, replace <\/span>\/\/ Saves the .pdf file to an actual location in the hard drive<\/span><\/span><\/code><\/pre><\/div>\n\n\n\nThe key thing I noticed when using the putpdf commands is that Stata would need the putpdf paragraph<\/strong> command whenever you start a new file\/page. But other than that, as long as you can keep track of your mental image of the yet-to-be-saved .pdf file, it’s pretty straightforward. <\/p>\n\n\n\nThe other thing is that you can see in this example that temp files can store things in whatever format as long as it is unambiguously specified.<\/p>\n\n\n\n
Now onto some cool stuff. Say we have a local holding a list of variables to loop through, and now we need to only loop through some of them. There’s actually a cool way to do this that will make you feel like a real programmer (if you’re a goofball like me, that is)!<\/p>\n\n\n\n
We are talking about list manipulations<\/strong> here. Keep in mind that all list manipulations work with macros and are supposed to work with macros only. Hence any locals should not<\/strong> be wrapped with single quotation marks. <\/p>\n\n\n\nSearching “macro lists<\/strong>” in Stata’s help file brings up the full list of things that can be done as list manipulations. I’ll offer two specific examples below to illustrate a bit of technical details:<\/p>\n\n\n\nStata<\/span><\/span><\/path><\/path><\/svg><\/span>\/* Example 1 *\/<\/span><\/span>\n\/\/ Imagine working with a long survey dataset where there are two observations (id=1 -husband\/2 - wife) per household (HHID). Here I'll reshape the hypothetical dataset to wide, and change the variable labels<\/span><\/span>\n<\/span>\nquiet<\/span> <\/span>ds<\/span> <\/span>\/\/ This the command that generates the r(varlist)<\/span><\/span>\nlocal<\/span> <\/span>varlist<\/span> <\/span>`<\/span>r<\/span>(varlist)<\/span>'<\/span> <\/span>\/\/ Exports a list of all variables<\/span><\/span>\n<\/span>\nlocal<\/span> <\/span>exclude<\/span> HHID id <\/span>\/\/ These are the two variables that do not need re-labeling after the reshape<\/span><\/span>\nlocal<\/span> <\/span>allvars<\/span>: <\/span>list<\/span> varlist-exclude <\/span>\/\/ allvars becomes a local that includes all the variables excluding HHID and id<\/span><\/span>\n<\/span>\nforeach<\/span> <\/span>var<\/span> <\/span>of varlist<\/span> <\/span>`<\/span>allvars<\/span>'<\/span>