9 hours 53 minutes
Welcome back, guys. I'm Katherine MacGyver in This is your lean six Sigma green belt. Today we're going to go over a graphical analysis.
So in our last module, we finally got to learn about these distributions that we've been talking about, really? For a course and 1/2 now because we talked about it in Yellow Belt and we said, It's coming. Um, and our last module, it was finally here, and hopefully
my intense for you is now all of the little pieces of information
start to knit together, So we're within our domestic framework were firmly and analyzed phase. Right now, one of the things that's always a little bit of a challenge for me for teaching
Lean six Sigma is you have your ideological peace and your structural piece, which is your PDC and your domestic models. And then you have all of these tools that could be used in multiple different places. Like, for example, what we're gonna learn today in graphical analysis can be done independent of a project.
It might be done early on in discovery when you're trying to figure out if this is even worthy of doing a project.
So during your defined phase. Or it could be something that you add as part of your control plan at the end of your control phase. So with that, I am hoping that all of the little breadcrumbs that I've dropped over the last two courses start to knit together. And let's go through your graphical analysis. So
remember our presidents
Remember Presidents Heights? We talked briefly about history grams in, um, yellow Belt when we started talking about this idea of how do we organize our data so that it makes sense on We talked about the Peredo principle, which we will talk about again in our next module
because it's very good for categorical data specifically. But one of the things that that's important is that
your history Graham, assuming that it is not sequenced by frequency like a Peredo diagram. But it is instead sequence by intervals like height and inches. This gives you
your distribution curve, so imagine we're overlaying of line,
and that right there gives you the shape of your distribution. So as you're looking through this and we've hinted around about normal distribution, I'm hoping that you are staring at the same. That is not a normal distribution because this is not a normal distribution with that. When we start talking about
normal distributions, we're talking about skewed nous or the proportion of data that is on one side or the other of the on mode,
not mode. Oh, God mean eso if you remember when I said that your mean your average, um you want your average and your median your midpoint to be, um, zero. What that is is giving us a sense of skewed in this.
So hopefully you have played around with your data analysis tool path
and Excel homework from last module, and you are familiar with the idea of skewed nous is it pops up into that table. So let's talk about what skewed nous actually means. Skewed nous is a shift. It means that it is not a, um, a normal distribution.
It is a newer term, and I say new return and the reason why I popped this graphic in here is Does this is how I learned it.
We call them tales. So in statistics, if you look at your normal distribution your bell shaped curve right in the middle with two puppy dogs, you have your two tails. This is going to be your trailing data. Remember, 68% gonna be within one standard deviation of the mean
90 98% giver, plus or minus is going to be within three standard deviations of the means. Those air those tales shapes like a puppet up when we see skewed nous, what we're seeing is that there is a larger proportion of numbers on one side or the other side off the mean So your average
is not, in fact,
straight down the middle. So when we say we're skewed left, you will also hear this as left sided tales or skewed right, right sided tales. I use the puppy dogs because this is how I remember the tail is on the left side. We have a skewed to the left. Keep that in mind. I'm a very visual human
with this skewed nous. Now the question is, what does it mean? Why do we care? This is going to be the behavior of your data. So we talked about central tendency on them. We talked about how we like data that is a normal distribution curve makes a perfect a bell shape.
But where we know on either side, we could determine probability.
If we have skewed nous in our data, our probability determinations go straight out the window. It tells us that we have some sort of variation. That is not the way that data perform. So back to our Quinn clicks, we know that dropping balls down, uh, graph gives us a normal distribution curve.
We know that if we look at large enough data set,
we know that there is a normal distribution curve in there. If there is not, there is something else in play that you, as the lean six sigma practitioner, need to use your root cause analysis your five wise your Shiokawa diagram your, um,
affinity diagrams to figure out what is it that is causing this your data
to perform in a non normal manner. This is really important. Skewed nous is so important for your CPK in your PPK. It is also extremely important when we start talking about statistical process control. In a few modules we talked about in our descriptive statistics model, we said, Hey,
the descriptive statistics is our fast analysis.
So if you're mean is less than your median, so your median is larger than your mean. You have a negative skew or left sided skew, which means that the majority of your numbers are going to be on the right side, so negative
pulling down towards zero positive. Conversely,
pulling towards a higher number symmetric that's a normal distribution means you're meaning your median are equal or really, really close to equal. So remember, sample size determines how accurate your descriptive statistics are, but larger samples air better
2020 sample points. 5% can change it,
but you want to look for mean and median as close to equals possible. This is good because it means that you can predict the behavior of your process. Positive means that your median is smaller than your mean. So what that means is you have this tailing effect towards positive numbers,
larger numbers. That's how we move back and forth according to zero.
So that right there is our fast analysis. When we talked about when I said that once you get to be a really, really savvy practitioner, you can look at these two numbers and say OK, you have a non normal distribution. That means there's something that we need to do a root cause and figure out congratulations. You are now a really savvy practitioner
because you can look at me and and median and get an idea of the shift in your data
so that we have a pop quiz. Um, if you remember, we looked at our distribution shape for the U. S. Presidents. Now I want to know from you guys, Is this left right or normal distribution? We're looking at our history, Graham.
All right, I am really, really hoping that you guys thought of the puppy dog like I did. And you registered that This is a left distribution or a negative distribution, which means that the majority of our numbers are going to be to the right of our median.
With that, that is the start of your graphical analysis. That's going to be a good big chunk of what you dio. So we talk about history rams in yellow belt because hissed a grams become the foundation that become your distribution. So you no longer a ranger data categorically
and like the Peredo, which we're going to do next.
But instead you arrange your data on that continuous spectrum. However you measure it and now you could start seeing how you perform. We also went over skewed nous. The important thing is, if you're mean in your median, do not equal zero. There is something in your data that needs to be drilled down into,
and you're going to use your root cause Analysis tools
from your yellow belt.
The next module. We're going to go over the Peredo parent principle in more death and how to build those graphs, so I will see you guys there.