What's the easiest way to converting raw numbers into a chart or graph?

October 20, 2007 6:09 AM

What's the easiest way to convert this ungodly mess to a nice chart or graph?

I have bilateral tendonitis and lately it's been painful for me to do anything on the computer. I'm looking for recommendations on how to parse these product names and numbers into a sane looking chart or graph with a minimum of keystrokes.

posted by JaySunSee to computers & internet (6 comments total)

You could try adding tabs between each "column," like so:

% change 
Trade class[tab]Unit sales vs. year ago[tab]Percentage change
 
TOTAL MASS MARKET ([dagger])[tab]12,300,850[tab]-21.9% 
DRUG STORES[tab]5,459,096[tab]-17.9% 
SUPERMARKETS[tab]3,640,303[tab]-26.8% 

and then saving it as a .txt or .csv file and opening it in Excel. Normally I'd recommend enclosing each entry in quotation marks and using commas instead, like so:

"Trade class","Unit sales vs. year ago","Percentage change"
 
"TOTAL MASS MARKET ([dagger])","12,300,850","-21.9%"

as I find comma-delimited files to be more bulletproof than tab-delimited files, but you did say fewest keystrokes.
posted by chrominance at 6:27 AM on October 20, 2007

That's a rough set of data to be massaging with a minimum of keystrokes. I'm assuming that it's the only form in which it's available, after which I'd use a whole whack of messy Perl to grab section names and parse the figures from the end of the line backwards. Since the product names have spaces and the lines you want are also space delimited, you'd have to map the structure of each section in the parser.
posted by rhizome at 8:31 AM on October 20, 2007

here's how I would visualize this - and it can easily be done with Excel.

each of the products is represented by large histogram blocks: one block for the Whitestrips Premium, one for the Whitestrips, one for the Rembrandt, etc...

These histograms' heights (eg each product's annual sales) values range from $2M to $26M so this gives you a nice visual range against which to build the next line graphs.

Then you build two line graphs in front of the histograms, contained within each of the product blocks:

One line graph for the dollar sales (for example, Crest Whitestrips Premium -24.4% 25.7% -3.5%)
repeat for each product, for each histogram block

Then draw another line graph in a different colour for the unit sales (for example, Crest Whitestrips Premium 858,158 -22.1% 15.7%)

So now each histogram block represents one product (for example, Crest Whitestrips Premium, $26,932,920 ) with line graphs for $ sales and unit sales for each product

You represent the product's market share as numeric information beneath each histogram block (Crest Whitestrips Premium for example will state -0.9% )

As for the information that you have on the top for the general tooth whitening product market, you can create a second graph and use 3 values - left for the dollar sales, right for the unit sales and list the trend beneath each block

Now that you know this, it's just a matter of setting up your graphs in Excel, and that will take only 5 minutes. the key issue here is organizing your data (and I agree with chrominance that CSV works best
posted by seawallrunner at 8:35 AM on October 20, 2007

Have you thought about just hiring someone to do it for you?
posted by SampleSize at 8:49 AM on October 20, 2007

The mess looks like it's been through the email wringer, and most of the keystrokes you need are going to be devoted to unwringing it. Is there a way to get hold of the original email file, complete with all headers, or better still, the file that was originally pasted into the email?
posted by flabdablet at 9:28 AM on October 20, 2007

Actually flabdablet, that's pretty darn close to how it looked on the original journal document.

Thanks for all your suggestions thus far.
posted by JaySunSee at 9:49 AM on October 20, 2007