Nov 11, 2012
I am running Excel 2003 on a Win7 system.
View 1 Replies
Here is my situation:
Each of my data sets spans roughly 75 columns by 250 rows at present, but this could expand. The first 7 rows contain metadata. Columns 2-25 or so contain the raw data, from which everything to the right is calculated. The data sets have most columns in common, but not necessarily all.
In order to tease out the most meaningful information from my data, I frequently sort all or part of it based on varying criteria. When I find a useful sorting criterion, I create a new column with a header that describes the criterion and populate it with a formula that returns a 1 if the condition of interest is met for that row, or a 0 if it is not. For example, if I am doing this in column AA, I might enter
and copy it down to the end of the data. The resulting vector of 1s and 0s quickly re-identify data that meets that criterion even after subsequent resorting. It also makes locating data that meets multiple sorting criteria extremely simple. Essentially, I create a truth table.
Cell $AA$4 in the above example contains a "comparator" value I might wish to change at some point, which would change the subset of data the condition selects for.
Here's the first hard part:
For each data set, I need the ability to generate meaningful plots that includes separate series based on the criteria I have described. However, I also need to retain the ability to resort the data or change the comparator value without disrupting these plots. In other words, the plots must NOT change when the order of the data is changed, but MUST change to display the appropriate data when the comparator changes.
Here's the 2nd hard part:
Once I have this working for one data set, I need to be able to port it to other data sets (which are contained in other workbooks), so that I can compare equivalent plots from each. I also need to minimize the number of manual steps involved in doing so, to avoid human errors and excessive time consumption.
The only other possible complication I can think of at the moment is that, to this point, I have been inserting blank rows to isolate subsets that I do not wish to perform further sorting on from each other.
Right now I am angling toward VBA code that loops through the entire data set to generate base dynamic ranges using the column header row (row 1) as the names, and the entire column of data for the rangeloops through the truth table columns to generate "branch" row ranges for each of the sorting conditions,loops through the entire data set one more time to create "branch" ranges for each of the base ranges.
I could generate some code to accomplish a one-off solution for a given configuration of a single data set (provided there is not a list length limit in a chart series that I'd be violating)...but without a dynamic named range, I don't know how to get to something that would update appropriately. So in essence, I am still stuck at the dynamic range part of this.