Analysis of an Outlier

A couple of things to mention before I get into the heart of this post:

1) It's that time of the year again where I have a lot more things to do and paradoxically I get close to nothing done. Like this weekend. I had planned to do quite a few other stuffs this weekend but then midway on Saturday, I realized that I have only 1 more week until Thanksgiving. Which means if I want to beat the traffic, I have to do all of my shopping before Thanksgiving. Last night I realized I was running low on personal supplies which meant I was going to have to shop for them after Thanksgiving. So today, I did the shopping which took most of the day. And I missed some stuff, so only one more weekend to do it.

2) My website is still having problems. It keeps giving me the 500 error, the edit page or whatever page I want to go to simply does not show up on my screen, or it runs very slowly. These issues are really going to cut back on my posting.

3) It seems every fall (except last year) I run into problems with my machine or with the website. It kind of feels like I've been hacked. It is like there is a little gremlin running around out there deciding to give me problems because it happens every fall (except last year - maybe they were busy with the elections).

[ap_divider color="#CCCCCC" style="dashed" thickness="2px" width="100%" mar_top="20px" mar_bot="20px"]

Now to the main part of this post.

Earlier in the week someone pointed out that due to the downed units at her site, the report that we do every week was showing strange numbers - they looked incorrect. Unfortunately, those incorrect numbers are part of the bug. The math was done differently on purposefully, and my boss and I came up with a possible reason why for doing the calculations in such fashion (we inherited the task of rolling out this report duty to our side of the business so we had to understand the use of this report AND the prior person was very emphatic about this particular calculation.)

Normally those calculations do not have an impact but we are now in an outlier situation where we have a lot of downed units due to the hurricane that passed us a couple of months ago. Now the numbers look really wickedly wrong. 

So this week I tried to do some analysis to see at what point I should have the macro show another tab with the "correct" calculations rather than the one we were using. In other words, I would have 2 report tabs: one with the "incorrect" calculation and the other with the "correct" calculations. Normally the "correct" calculations tab would be hidden until a certain threshold of downed units is reached at which point the new report would then be offered for distribution.

What I was trying to do was figure out if there was a common point where the variances between the "incorrect" calculations and the "correct" calculations become too material. This would be the point where the macro shows the "correct" calculations. I could do the analysis by varying the number of down units so I could see how the results played out but I wanted to have something do it automatically for me because I wanted to play it out for various sites. The idea that came to mind was to use the data table to display various scenarios. Unfortunately, the report's calculations was a bit more complex so I was forced to manually try different down units.

On Friday afternoon, I tried it again and got further until I ran into a problem. This morning a possible solution came to mind but I had to spend the day shopping. Hopefully the solution works and I can get to the point where I'm actually using the data table method. And then I want to try to apply something about the data table that I had read some time ago. What I really want to end up with is a method of analyzing situations so I see at what point things go weird...if there is a grand picture. It may end up being a "nothing burger" - there is no common point at which numbers go weird or becomes problematic.

Sorry, this post is a nothing burger post too. I think from here on until sometime January, my posts will be either sparse or "nothing burgers".