Comments on: Using the MATCH Function for a Linear Search and a Binary Search

By: Chris

Chris — Tue, 06 Apr 2010 23:07:00 +0000

I have to say iframes are almost as annoying as pop up ads and pop under ads. They mess with mouse focus and depending on the skill of the site author they often are large than the monitor so you are forced to repeatedly scroll two different places to see everything. For my money (aka time) I just skip away from iframed content onto the next thing I am interested in reading – too much effort to read!

By: Mathias

Mathias — Thu, 18 Mar 2010 23:27:00 +0000

@Dick: thank you for clarifying my comment, this is exactly what I meant!
@Tushar: sorry if my comment came out sounding aggressive, this wasn't my intent! I was just pointing out what I thought (and still think) is an issue with the publishing solution you were describing.
Re: the discussion about how much to post on a feed, I understand (and respect) why some authors would just post an excerpt on the feed, if driving visitors to the actual website is of value to the author. That being said, if I can't see any part of the contents in my feed reader, which is the case with this post, you might as well kill the feed, because it serves no purpose at that point...

By: JP

JP — Tue, 16 Mar 2010 15:22:00 +0000

@Jon — My feeds have whole posts in them, but I could see how some sites (especially those under constant attack) would want to protect their content by showing only excerpts.

By: AlexJ

AlexJ — Tue, 16 Mar 2010 04:31:00 +0000

Thanks Charles – I reviewed your DecisionModels site – I think my sin was doing matches off-sheet. Now for my penance…

By: Jon Peltier

Jon Peltier — Tue, 16 Mar 2010 03:48:00 +0000

Jimmy -

It's a balancing act between protecting your stuff (from scrapers) and sharing your stuff (with readers). I err on the side of the readers, and once in a while track down the scumbags that repost my stuff.

By: Charles Williams

Charles Williams — Mon, 15 Mar 2010 21:09:00 +0000

Its possible to make a linear search UDF that is faster at recalculating on unsorted data, by storing the row number that was found in the first calculation and then checking the result for that row number on subsequent recalculations.

This algorithm is only faster if the data to be found exists in the same place as it was at the last calculation.
So for example if you add a row at the top of the data it won't be any faster, but if you change a single value in 10K rows of data it will mostly be a LOT faster.

for most spreadsheets this works well in practice.

By: JP

JP — Mon, 15 Mar 2010 20:43:00 +0000

Some blogs only publish excerpts in their RSS feeds to prevent scraping.

By: AlexJ

AlexJ — Mon, 15 Mar 2010 20:24:00 +0000

Tushar: Thanks. It appears that COUNTIF is no faster. I wonder if a UDF might perform better.

By: Tushar Mehta

Tushar Mehta — Mon, 15 Mar 2010 06:02:00 +0000

Alex: If the performance of your Excel model is acceptable, I wouldn’t change anything. As far as COUNTIF goes, I suspect it is slower than MATCH in linear mode since it must check every value whereas MATCH can quit as soon as it finds one match. But, you may want to verify the results for yourself.

By: AlexJ

AlexJ — Mon, 15 Mar 2010 04:03:00 +0000

Tushar,
I've been using exact match to determine of a value exists in another data set (=NOT(ISERROR(MATCH(SearchValue,SearchData,0)))

Based on your article, I should use another method. I don't think I should use the Binary seach since SearchData is not sorted which would give unreliable results. My next best idea is =(COUNTIF(SearchData,SearchValue)>0). Is COUNTIF any faster than the LINEAR unsorted Search?