[quote user="dishdy"] I think you can safely assume YYYYMMDD. In fact, if you look at the convert function parameters, whenever a date starts with YYYY or YY then it is always followed by the month. Thus assume convert(varchar, ... ,112). [/quote] Dis

Sorry to be a pest Jacob, but one more question if I may? The sample data lacks sufficient difference in the output data for me to tell whether we mean to show transactiondate as YYYYMMDD or YYYYDDMM. The former makes more sense as it is more widely used,

Thanks Jacob!

Hi folks, quick question re. the sample data for challenge 33. On the webpage, the eighth row of data is shown as follows: AuditId BookingId TransactionDateTime TransactionType TransactionValue 8 3 2010-01-01 19:00:00 R

I found this almost suspiciously straight-forward, but that may be because the products I work on deal with this sort of issue all the time. From my experience though, I know that my method probably won't scale very well for large numbers of phrases and/o

It's a simple way of displaying a set of values in such a way that their relative magnitudes are most obvious. Here's an example of how the random unsorted data would look (using the challenge's example data): And here's the data post-transform:

In response to dishdy's and uzariahm's comments regarding NULLs: To put this challenge into the real world isn't impossible, and in doing so it becomes apparent that NULLs could easily occur. Imagine that the 5x5 matrices are actually the votes from