ASQ Six Sigma Green Belt – Objective – Ethereum

1. Ethereum Overview

For Six Sigma, you use DMAC approach. And once again D in DMAC stands for define, m is for measure, a is for analyze, I for improve and C for control. So far in this course we have talked about defining line and measure phase. So now the next step is to analyze that data which you have collected in the measurement phase. In analysis of data, there are two broad things you could do. One is explore the data, look at the data from different points of view and make some sense out of that data. The second step which you could do is to conduct hypothesis testing. So here you check whether the mean has changed, whether the variation has changed, what is the thing which is affecting the process. So that is something which you will be doing in hypothesis testing.

Now let’s come back to exploratory data analysis. Here you are exploring data. You just look at the data and see if you can make any sense out of that. Once you make some sense out of that data, then you can do some statistical tests to confirm whether your assumptions are right or wrong.

So those tests are covered in hypothesis testing. But right now what we are doing is we are exploring data, looking at the data and see if we can make any sense in exploratory data analysis. Also we have two things multi vary studies and correlation and regression. Let’s start with multivariary studies. In multiwarry studies we will be looking at three broad topics here. One is the types of variation. Because Six Sigma is focused on reducing variation, bringing consistency. So we will understand what are the three types of variations.

Then I will take two examples. The first example will be related to bearing measurement. So I have a bearing and I take measurement for the diameter of that bearing. So that is one example I will take. And the second example I will take is for the call center. For second example I will use minitab. Let’s start with three types of variation.

And before that let’s understand what is multivarry studies. So multiway chart, which is the outcome of multi vary studies. So this chart is a tool to visually show a variety of sources of variation. So whatever measurements you are taking, there could be number of sources of variation. And this is what you want to find out. What are the sources of this variation? Once you find out the source of the variation, then you can act upon that to reduce variation.

So here we have three types of variation and these are positional, cyclical and temporal. Positional is related to the position or the variation within a part. And once I take the example of that, ball bearing. So there you will have much better understanding of these types of variation. So there I will be talking about positional and cyclical. Cyclical variation. Positional is within part, within one bearing and cyclical is between parts, between various bearings. And then the third type of variation is the temporal variation or the variation which occurs over a period of time. This is something which we will be talking in the second example, when I will be talking about the call center.

2. Ethereum EVM

What I’m doing in this example is I have ball bearing production line, from which I pick these ball bearings and measure the size of that. I need to control the outside diameter of these ball bearings. The nominal value for this diameter is 52. 0. This is what you see on the picture on the right. Now, what I do is I pick three ball bearings from the production line. And on each of these three ball bearings, I take three measurements, three measurements in different direction, in different locations. Let’s say I take ball bearing number one, and I take three dimensions for that, dimension number one, dimension number two, and dimension number three.

These come out to be 52. 0, which is perfectly what I wanted. The second one comes out to be 52. 2, slightly more than the nominal value. And third one is 52 point. So this was for ball bearing number one. Then I take the second ball bearing, and I do the same thing, take three measurements for that, and I take the third ball bearing, and I take three measurements on that ball bearing.

So here are nine values which I collected by measuring these three ball bearings. Now, I am interested in reducing variation, because what I want to do is I want to produce very consistent diameter for these ball bearings. There is a variation in the diameters which I can see here. But then where is this variation coming from?

This is what multi vary chart will tell us. So with these nine measurements, what I do is I use minitab, and I draw this multivarry chart. And how I draw this multi vary chart, I will talk in the second example when I take the example of call center there, I will be using minitab to understand how you plot multi vary chart here. But for now, let’s understand that this is the chart which you plot. So what does this chart tell you? So here is the size 52. 0. This was the nominal value. And these are things on plus side. These are values on the minus sign.

This is ball bearing number one, ball bearing number two, ball bearing number three. And here I have three measurements, measurement number one, measurement number two, measurement number three. For ball bearing one. Similarly, for ball bearing two, I have measurement one, two and three. And so on one red line, which you see here, these are the mean of ball bearing one, two and three, the diameters of ball bearing one, two and three.

So now if I just look at the mean and probably without knowing multi vary chart, this is what I would have done. I would have taken ball bearing number one, took three measurements from that, and took the average of that as the measurement for ball bearing one, for two, and for three, I would have done the same thing. And this is the graph. I would have plotted that on ball bearing one, the average diameter is around 52. 0. For ball bearing number two also is around 52. 0.

But on ball bearing three, I have the average diameter, which is less than 52 point. So three is smaller size. That’s all I would have concluded without knowing multivarry chart. But what multi vary chart does is this introduces multiple dimensions in the chart. So in addition to the average of ball bearing diameters, this also tells you the positional value, the value of diameter at three different positions. So this is what you would call as the positional variation. Here, if you see the variation within ball bearing one, this is the positional variation variation. Because we are taking measurements at different position, in ball bearing two, we have a lot more positional variation.

The second type of variation, which is the cyclical variation, cyclical variation is the variation between these items, between these ball bearings, and that is represented by the average. So when I look at between these ball bearings, I could say that ball bearing number three is of smaller size. But if I look at the within or the positional variation, I could see that there’s a lot of positional variation within one ball bearing itself. There is so much of variation that between ball bearing variation really doesn’t matter. Ball bearing number three is small in average diameter. But if you look at these values, you really cannot say with very confidence that ball bearing number three is smaller size because the variation within the ball bearing is too much, or the positional variation is too much.

So this is how you look at this data. Now let’s take another example. Another three bearings from another production line. So what I have here is a separate data which is m two. So this one is measurement number one, m one, which you can see here. Now I take another measurement, which is measurement number two, which is from machine number two.

So here I have data from machine number two. So from there also, I pick three ball bearings, take three measurements for each of these. This is what I get. Now, if you look at multi vary chart, you will see that the variation within ball bearing is not that much what we saw earlier, and the variation between ball bearings is more. So here we can say that ball bearing number three is definitely small in size, two is definitely higher in size, and there is very little or small amount of variation within these ball bearings.

Now what I will do is I’ll put both of these charts, the chart from machine one and machine two, side by side, to have some better understanding of the behavior of machine one and machine two. So here I have two charts. The first one is for machine number 1. Second one is for machine number two.

Now, putting side by side, I can very well say that the first thing which I need to do is look at the variation within ball bearing for machine one. Machine One is doing something because of which we are not getting the consistent diameter in individual ball bearing. So ball bearings might have ovality or something is happening so that the ball bearing is not consistent in the size. Forget about the variation between ball bearings. So this was the first example where we looked at two types of variation positional.

3. Ethereum Gas Estimates

Here I have my minitab session and the file which I have opened is the call center file. This file is available in the resource for this particular lecture. So you can download this file if you want to play with this using minitab. So let’s see what data we have here. So we have eight columns here. The first column is the call center. So there are two call centers here.

The first one is Monitor Paler and the second one is the St quent time. My pronunciation might not be perfect for these names, but these are two call centers here and in these call centers we have number of requests. These requests could be related to credit card. These requests could be related to individual accounts, queries or some technical thing. So all these are the type of requests which you get in the call center. Then this is the date on what date this call came. And this is the duration, how much time it took to attend to that call.

And this is something which we are interested in knowing that what all are the things which affect the duration of the call. And then we have customer categories. So there are different customer categories, ABCDE, these could be different type of customer which could be existing, new or high value customer, low value customer, whatever it is. But there are five types of customers here. Then this is the week number and this is the hour at what time and this is the day on what day this call came.

So this is the data we need to analyze. And what we are doing here is exploring the data because in analysis we are in the exploratory phase. So we are exploring the data to see what sense we can make. So if you look at this data, this data has 980 rows, so a lot of data. But then just looking at this data doesn’t make any sense. So we need to make some sense out of this data.

And that’s where we are using multiwary chart. So for that I go to Stat and I go to Quality Tools and I click on multivary chart. In multiwary chart, what is my response? My response is the duration, the time it takes to attend to the call. That’s what I am interested in. So in response I will double click on duration. That’s my response. So the first thing which I want to do is I want to see whether there is a difference between call centers. So I have two call centers, whether there is a difference in the response time for these two call centers or not. So what I will do is in factor one I will double click on call center. I can add up to four factors here, but let’s take one factor only. And with this if I press OK, this gives me a chart. I will not call this as multi vary chart because this doesn’t contain multiple sources of variation. This has just one source of variation, which is the call center. So the average for Montpellier call center is 588 minutes. Same thing if I look at St. Quentin here, the average is 525, which is significantly less than Montpelier. So this is something which comes to my attention that there is a significant difference in the time it takes to attend the call. That’s my first conclusion.

Okay, so I need to go a little bit more further into this because right now I just look at the call center. Now, if I want to look at what are the types of requests these call centers are getting, and there what I need to do is I will click here, which is the edit last dialog box. So this brings the old dialog box. So I don’t need to double click on duration and call center. Again, what I need to do is I can add another factor, which is the request, the type of request. And here I’m just picking few things just to understand the data and I don’t put the third factor here. And with this, if I press OK, this gives me a multi vary chart related to five different types of requests.

One thing which I can very clearly see here, that technical support takes a lot of time. So what you have in this multi vary chart is the type of request. And for each request, what is the time for each of these office. So the line in the red is the average for both of these call centers. But then you have individual value of these call centers as well. So if you look at this data here, you see the average and you see one call center which is Mont Pierre. So this type of query, which is credit card query only comes to this one office only once it comes to individual accounts. So I have two call centers here and in the middle I get the average of that. New accounts only come to one call center, which is St. Quentin. Queries also come to one particular call center which is St. Quentin, and the technical support come to Montpellier. So one thing I can see is one particular office which is Montpellier, gets technical supports and it takes a lot of time to resolve. So all other types of calls take around 500 to 550 minutes. But if it’s a technical call, that takes 750 minutes.

So if I put these two charts side by side, that gives me some idea. That why this particular office, which is Montpellier, takes more time on the average, because this is the only office which gets technical support. And technical support takes a lot of time to attend to, which I can see here in this particular multivariate chart. Okay, so let’s take this investigation a little bit further. And here what I want to do is I add another factor, which is the day. So let’s do that. Click on the last edit dialog box and add the third factor as the day of the week as well. The day is something which is related to the temporal variation or the time variation, time related variation. So if I click on this here, I have the third multivarry chart, which is getting a little bit more complicated here. So let me make it full size here. So this multivarry chart gives you the time.

Monday, Tuesday, Wednesday, Thursday, Friday, Saturday, Sunday. And this also gives you the five types of queries and it gives you the office as well. And these green line gives the average of everything. So first thing which I see here is that on Monday it takes more time because of whatever reason it is, any call which is coming on Monday generally takes more time. And you see the peak here, this peak is for the technical support. So technical support takes more time any day. So this is how you use multiwary chart to investigate your data. Let’s quickly do one more thing before we close this session. And here what I want to do is I want to look at the call center versus the day. So what I will do is I will keep the response as it is. Call center is something I want to see. Let me delete the request.

And instead of request, I just want two variables, call center and day, because the previous chart was becoming too complicated. So let’s make it slightly simple and see how much time it takes by call center on a particular day. So here is another multi vary chart which tells me that Monday takes more time on the average. And these are two offices, montpellier office and St. Quentin office. And this is the average on Monday. Same thing on Tuesday, Wednesday, Thursday, Friday, Saturday, Sunday. And every day you will see that every day Montpellier takes more time. And we have talked about that earlier as well, that more time is because of one particular type of query, which is the technical queries that take more time. And that’s the reason Montpellier takes more time. That particular call center takes more time to attend to all the calls.