Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Locked thread
Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!

AJzer posted:

Hey! Programmers over in the Cavern of COBOL have no idea what I'm talking about when I discuss deep excel financial models! Go figure...

So here's some questions for anyone using Excel for complex modeling: What do you do when your model gets too big for Excel? Specifically, I'm getting into some ridiculous lookup tables, nested conditionals 3 deep, and 100x5000 cell-count worksheets (x6) that just need to be pushed into a database.

So what kind of solutions are out there using MS products and/or free ones? The theoretical underpinnings for my needs are pretty basic: linear regression and simple algebra, so Excel can handle the manipulations and reporting just fine. It's breaking every formula when you need to make a small change, and waiting 30 minutes to load a Workbook that I'm hoping to change.

I'm more proficient with the theory than the tools, but I'm not afraid to learn some scripting (Powershell, possibly VB), basic SQL or Access use, etc. Any wonks out there have suggestions?

What version of excel are you using?

Adbot
ADBOT LOVES YOU

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!

f2a posted:

Looked at MATLAB? Failing that, Python with Numpy+SciPy is a very good (and free alternative). Both have more of a learning curve than Excel, but you get a lot more power and flexibility.

The alternative is VBA, since you can shove it behind excel and integrate with your existing work. It's probably less of a step than resorting to numpy or the like.

Matlab is great, though. :h: Matlab.

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!

tolerabletariff posted:

Work/life balance, yo

I thought that term had no place in this thread.

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!


Holy crap I had no clue you guys worked on what is basically a 50% total package bonus system. Does everyone get some form of bonus regardless (hence the bands of $$), or are these M&I figures quoted only given to select top performers within a firm?

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!

I really wish I could see actual examples of what you kids do in Excel. I want to know what sort of complexity your stuff gets to and how much you have to do.

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!

Its Miller Time posted:

Who has a personal email that isn't firstlast@gmail

I do because it's my high-school era account but I also have a first.last@ISP which I use for anything semi-professional.

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!

I hope you don't use the mouse in Excel whatsoever. Anything else would shatter my dreams.

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!

bam thwok posted:

SumIf. I avoid pivots whenever possible, since you'll get occasionally bullshit around source data changes, refreshes, missed check-boxes when recreating views, etc.

sumproduct((range1=criteria1)*(range2=criteria2)*etcetcetc) if you need to sumif across multiple columns

pivots feel like cheating, really

:excellyfe:

e: and if you use vlookup rather than index/match you are literally subhuman

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!

bam thwok posted:

I live on the wild side. I don't even use ",false" in my vlookups.

You're the reason the GFC happened.

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!

FunOne posted:

Why not just use SUMIFS?

EDIT: Here is what I do, define a named range for the data and for each column. Remember, named ranges can take ".", so I use that to create a poor-mans name space for my data. Then it is =sumifs(table.data, table.year, 2013, table.location, "HATS", etc....). Quick and easy to change the criteria to references and create the equivalent of a pivot without all the icky-pivotness. And supposedly even over selected criteria like that are supposed to be faster than array based functions.

Because most of my company still has office 2003.

Also, SUMIFS works on a boolean AND function between all of the criteria whereas you can use SUMPRODUCT to do AND and OR boolean functions between the operators. It's more flexible because it's an arbitary mathematical function rather than assumed to be a summation of a bunch of criteria being true or false.

take for example, SUMPRODUCT(volume)*((shape="Cube")+(shape="Cylinder"))*(height>100))

That'll sum the volumes of all objects with a height of over 100 that are either Cubes or Cylinders. It distills each criteria down to a 0 or 1, so you're able to chain them together in boolean maths in any way you like.

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!

if it doesn't have spaces in the list of data, indirect("A"&count(A:A)) will do that.

Modify the A's to B's if it's column B, etc.

Adbot
ADBOT LOVES YOU

Nam Taf
Jun 25, 2005

I am Fat Man, hear me roar!

More excelchat because this blew my mind as I'd never seen it, despite basically living in the program:

=SUM(Sheet1:Sheet5!A1) works as you'd expect.

We need a PYF excel tricks thread.

E: it doesn't even need sequential numbering in it. It treats the sheet order in the bottom-left as a sequential dimension. So you can do London:Moscow!A1:B5 and select across A1-B5 in all sheets between London and Moscow, as per the sheet order.

Nam Taf fucked around with this message at 08:31 on Sep 25, 2013

  • Locked thread