Скачать или смотреть ChartSmartXL What is a Performance Chart?

www.chartsmartxl.com
www.turningdataintodollars.com


This is Jay Arthur. Let's take a look at what is a Performance Chart. Many people ask me why do we need this level of specificity? Because without it you're not going to be able to actually improve and sustain your improvements. It's just not possible.

So, what is a performance chart? Well, normally we start out by tracking errors, defects, whatever measurements of some kind and from that, we can create a line chart. But a line chart is probably the dumbest chart on the planet. It can report what happened, but it can't tell you anything about how to improve, how stable is this, and if this is good or bad.

Oftentimes, I see companies that have sales figures and the sales you know go up one month and they all go yay, and then the next month are down, they go bummer what did we do wrong? It may be that you did nothing wrong all right.

So, what we want to do is from this, we want to calculate an average. This gives us something that's known as a run chart, and a run chart still isn't very bright and very smart. It was designed to be simple, so that people could do it quickly, but it's just not all that useful.
One of the things we can do is look at how far is each one of these points away from the average, and this variation allows us to then statistically calculate what's called an upper and lower control limit. Those control limits help us understand are we statistically and in control and performing predictably or are we out of control?

There's some rules we're going to apply based on this to look at our data so we can use individual data like defects per day or any other measurement that we do individually or if we were in manufacturing we might measure 5 samples at a time and see how they vary.

Then, from that we can calculate the upper and lower control limits so there's different types of charts for different types of data. So, here's our performance chart and you can see here we have what are called 1 Sigma 2 Sigma and then the 3rd Sigma level in here. These are called zones and the first zone called Zone C, 2/3 of your data should be inside of this zone.

If you think about a bell-shaped curve, most of your data should be in the center.

Then there's Zone B, 95.5% are in here, and Zone A, 99.7% of your data should be within the upper and lower control limits. These things help us determine if something is out of control.

If you have one point above or below the upper and lower control limit that should only happen 3 times out of 1,000. If you only have 20 data points for it to happen once is highly unlikely that might be some sort of anomaly to go investigate.

We could do some root cause analysis, and then 2/3 above or below 2 standard deviations. That's add 2 out of 3, that's 2/3 that shouldn't happen right because 2/3 of our data should be in the center in Zone A.

That's unlikely as well, and 4 out of 5 above 1 Sigma, well that's 80% or above or below the center; so, that's very unlikely again. These things happen 3 times out of 1,000 and so these invite us to go investigate.

Chart smart will make them read so you can see them easily. You may also have 8 points in a row above or below the center line, and this is what's called a run. So, if our process shifts at upward or downward, 8 points in a row above or below the center, then that means our process is shifted and we need to investigate.

Those are some of the rules and those allow us to then analyze and really understand what's going on in our process. So, once you've made an improvement, if you start to see one of these changes, then you need to investigate and figure out what you need to fix.

This is the only way to monitor your process and really make sure it stays performing. You expect it to perform another type of thing that can go on is a downward trend so six points in a row descending or ascending. This is another type of thing that’s statistically unlikely.
That's the essence of our performance chart. The center line and the control limits help us determine if our process is performing as we expect it to and to detect when it starts to shift so that we can actually take some corrective action.

These charts went to college and took statistics so that you do not have to. That's a performance chart!