APLawrence.com -  Resources for Unix and Linux Systems, Bloggers and the self-employed

Handling missing data


2006/12/14

I have an old Perl project that goes out to a Government web site, ftp's some files, massages them in various ways, and spits out some output. Over time, the project grew, and now does more than it used to. To keep it generic and simple, let's pretend that originally it went out and got "temperature" files every hour. That was simple enough: about the only error condition would be the inability to get one or more of the files it wanted. My program simply kept track of what it actually got and only ran the rest of the process on new files.

The section of code that does that isn't hard:


    foreach(@file_to_get) {
    $newerr=0;
    $ftp->get($f,$f) or $newerr=1;
    push @gotfiles, "$f" if not $newerr;
    }
    
       foreach(@gotfiles) {
          ...
    }
 

Well, then "precipitation" was added, and then each file might have processing for different "customers". A bit of extra looping, no big deal.. but sometimes we'd get a "short file" output.

Examination of logs revealed the problem: every now and then a "temperature" file would be unavailable at the ftp site, so the final output only contained "precipitation" data. From my point of view, that's perfectly reasonable: the data wasn't available at the input, so it's not in the output.

But the customer doesn't like this. "Fix it", they said (well, much more pleasantly, but effectively that). So the question I put back to them is "How?". In this situation, there are really several possibilities: we could:

  • Leave it as it is now. I guess that doesn't count toward "fixing it".

  • Don't publish the file if it is "short". That ignores the data that we were able to get; but if the ultimate result is pointless without ALL of it, then this makes sense.

  • Fill the missing slots with data tagged as "bad". That might be reasonable too, because sometimes data we do get is tagged that way already.

  • Use an older "temperature" file - these things get published every hour, so we could produce a complete file using older data - I don't know if that makes sense to the final consumer, but we could do that. If acceptable, the next question would be "how old is still acceptable?"


Programmers have to deal with stuff like this any time there is a possibility of missing inputs. I talked previously about missing fields in a tab delimited file at Handling missing data in inputs; this is a slightly different example.



Got something to add? Send me email.





Increase ad revenue 50-250% with Ezoic


More Articles by

Find me on Google+

© Anthony Lawrence



Kerio Samepage


Have you tried Searching this site?

Unix/Linux/Mac OS X support by phone, email or on-site: Support Rates

This is a Unix/Linux resource website. It contains technical articles about Unix, Linux and general computing related subjects, opinion, news, help files, how-to's, tutorials and more.

Contact us





I am fascinated by religion. (That's a completely different thing from believing in it!) (Douglas Adams)

Computers have been taught to distrust each other and will reject attempted connections most of the time. Nowadays, most computers and firewalls are utterly rude about it: it would be like asking someone to dance and having them ignore you as though you were invisible and inaudible. (Tony Lawrence)












This post tagged: