## You’re “Poisson” Running Through My Veins: A Truly Epic Guest Post on Call Centers and Erlang C

“I Want to Taste You But Your Lips are Venomous, PWAH-SAAHHHNNN!!!!”
(Get It?  Poisson/Poison?  OK, Read on for a Bell Biv Devoe Reference)

### Intro from Rob

Um, wow.  A few things:

1. Brace yourselves for a dose of awesome.
2. I don’t understand everything that’s happening in this guest post.
3. So if you “get” all of this, fantastic.
4. If you don’t, don’t sweat it – just bask in the power of our toolet – it can truly do anything.
5. Our new friend Josh is absolutely killing it with his song references.

### Take it Away, Josh…

Since taking on a role in Work Force Management about a year ago, I’ve learned one thing: Staffing a call center is expensive. What I mean is: the staffing software, it’s is rather pricey. So much so, that smaller call centers just can’t afford the tools needed to easily create an accurate staffing model.

But as someone raised to the mantra of: “if you are going to do something, do it right” I decided to learn me some DAX. (To be fair though, what my dad really said was: “Aim low, that way no one can tell when you fail.” But for the sake of this post, we’ll go with the first quote. )

Luckily, Rob was nice enough to teach us the core of using complex equations in his Experiments in Linear Regressing, Parts 1 & 2. So we won’t be entirely lost in new territory, it’ll be more of a: “lost with friends and colleagues, ‘Danger Will Robinson’” sort of excursion.

### Using RankX and SumX to create a weighted moving average

The staffing model I use relies on a weighted average of the 4 most recent weeks of incoming calls. Often times however, a week’s data may have been inaccurate, causing us to go a week further back.

The way a weighted average works is that each number is multiplied by the given weight and then divided by the the sum of all weights. So the weights 40, 30, 20, and 10 are assigned to the weeks, giving us an average number of calls that is more heavily influenced by the most recent week.

#### The wrong way to do this:

I include it here because the interactions between the eight weight measures are really, really neat to watch.