In this lesson, we continue our example of Part 1, and Part 2 of QuikSigma Design of Experiments Part 3. This will greatly increase the efficiency of your Six Sigma projects. Have any questions? Leave them below and we’ll get back to you!
So here we are, back at the simulator, to get some data to analyze in QuikSigma. Now, you may recall, but what I did as I stripped out the last five variables, through H, and so all we have to pay attention to in the coded units matrix are columns A, B, and C and likewise over here. D through H show up but that they’re all frozen at the same value so they’re not going to have any effect. Now, the other thing I’ve got over here is every time I push F9, I will get 5 examples of my output variable that were taken at the set up that was specified here in the matrix. So when I set A, B, and C in their low state and push F9, I get 424, 530, 512, 495, and this’ll be off screen for you, 469. Well, we only have four replicates to populate and I was very careful to do that and was going to show you that I had exactly the same data in QuikSigma as I’ve got here, and lo and behold I pushed F9 and I can never get those numbers back. So you’ll just have to believe me that the numbers put into QuikSigma are these again. So let’s take a look at what they are like.
Once we put them in QuikSigma so here, we have our data set and if we’re looking to reconstruct what occurred in the simulator, I just read across the first row where everything was in its low state. I got a combination of that here so here was my first data and then here’s another set in its low state, that was the second column in the first row and then on down here, I got another combination where they’re all in their low state and that was the third one and then number 25 would be the fourth one in row 1 and similarly I populated the rest of this so we can have a look at it. Now, let’s then go here and click calculate and we did have a look in the data mining module and we can use this to interpret or we can simply go up here in the model and look for the low p value and it looks like variable C, which we arbitrarily called catalyst looks like it’s the active variable.
Now, there is one other thing that we can do and probably should and that is we should allow interactions. Interactions do happen and I’m going to allow up to two-way interactions. Three-way interactions are fairly rare. We could have one, we’ve got three input variables but let’s just take a look at what we’ve got here and we find out that the time and pressure interaction is not significant nor is the time and catalyst or the pressure and catalyst interactions. What’s an interaction? An interaction is when you get a result that cannot be accounted for by just linearly adding the two variables. Medically, if they give your blood pressure medicine, that lowers your blood pressure. They’ll also give you usually a little bit of a diuretic. That will have the effect of lowering your blood pressure but the two together achieve more result than you would expect and that’s an interaction. So I can click auto reduce and QuikSigma has a very smart way of going in and doing an iterative approach. It’s a very sound solid procedure and here we get it reduced down to the one variable that seems to have any effect.