738924_en-US

liliconcom

Industry Innovation: Blending Modern JMP/Classical Six Sigma - Applied Engineering Statistics (AES) - (2023-US-30MP-1413)

A leading semiconductor manufacturer has developed a novel multidisciplinary program for applied engineering statistics (AES), by incorporating applied problem-solving (6S DMAIC), JMP statistical thinking (STIPS), and core JMP curriculum course content, sponsored by JMP Education. The program's key features culminate in the use of JMP 17 features in powerful, systematic, and practical analytical contexts. The 6S Black Belt curriculum has been seamlessly innovated, based on both AES and JMP curriculum content across DMAIC phases.

The first initiative involved transitioning from Minitab 19 to JMP 16, followed by mapping the JCSA, JGPH, JANR, JMSA, JSPC modules to the DMAIC framework. These JMP core courses help facilitate DMAIC Black Belt Project execution more effectively through AES thinking. Several JMP case studies are demonstrated, including item analysis (Attribute GRR), text mining/data mining hybrid correlation, model-driven multivariate SPC, and group orthogonal supersaturated design.

To train internal instructors to teach Black Belt JMP modules, several advanced JMP curricula (JMP 17) have been created. In addition to internal Black Belt certification, the JMP STIPS Certification Exam was offered, as was the JMP DOE Certification Exam. JMP STIPS courses were made available to any internal employee. The program innovates the traditional 8D project framework by adding JMP 17 tools to make 8D project root cause analysis more objective and data-driven to resolve complex system-level quality challenges. The 2023 March Madness Forum Events, sponsored by 64 presenters, have drawn full attention across the organization, recognize the success of the JMP-oriented AES program.

All right , well , thank you everyone for joining us .

I 'm incredibly honored to be here for JMP US Discovery Summit .

I will be presenting the project titled Industry Innovation Blending Modern JMP

and Classical Six Sigma Applied Engineering Statistics .

This is otherwise referred to as a Lean Six Sigma JMP Based

Black Belt Program , which is a novel program

really championed by Charles Chen , my co -author and co -presenter .

Let 's dive right into it .

First of all ,

what is the high -level roadmap for this program ?

It really starts with transforming a global quality culture

and ends with connecting local JMP SMEs and Master Black Belts .

The way that we 've done it is through a segmentation of programs ,

A plus , A minus , and A .

Really , we 're focusing on the A plus program

being the elite program ,

and we 're going to get into that much more .

But the outcome is really for a focus on the entire global quality initiative .

This program covers a 2 -3 year span ,

and it 's really based on a 2 -3 year outcome .

But the ultimate convergence of it is that this organization use case is that

we 've decided to use JMP for all the programs ,

including the Six Sigma program , and always with the intent to deliver it

with the highest quality in alignment with the global quality strategy .

A key feature of this program is encouraging healthy competition

through hosting JMP forum events , which we 'll showcase a little bit .

What we 've shown here is that we 're using a Six Sigma tool

called the SWOT Assessment, s trength , weaknesses , opportunity , and threats,

as you can see in this matrix .

The things I want to emphasize here are that

using JMP as both the external curriculum

and the internal curriculum is fundamental to this program .

What I mean by that is ,

which I 'll show later , is that we 've really combined

JMP education 's curriculum ,

which they 've provided to the customer base ,

for a nondisclosure agreement with an internal Six Sigma curriculum .

An applied statistics lean with JMP ,

plus a rigorous Internal Six Sigma applied engineering program .

It 's important that our leaders focus on opportunities .

This methodology really focuses on building details ,

but if we go from details up to the nucleus ,

we 're talking about leaders who focus on opportunity .

We want to start small , think big , and act fast .

This is one of our main tenets .

On the opportunity side ,

really looking at how we can synergize the JMP program , the JMP 16 /17 curriculum

with the BB curriculum , and create an optimum recipe

for leadership development , learning, and deployment across the organization .

The other thing I want to mention here is that this slide

highlights a migration from Minitab .

While it might indicate a shorter -term productivity loss ,

it 's a huge return in medium to longer -term productivity gain .

As everybody probably who attends Discovery knows ,

JMP 's interactive graphing capabilities and multiplatform interaction

and flexible and powerful statistical modeling capabilities

with a scripting engine and JSL really make JMP far more powerful

than Minitab in today 's analytics era .

This is a key milestone for us that we achieved early on .

How does this A , A plus , A minus tiered program design work ?

Basically , we 're using the three levels

to effectively segment our trainees , our stakeholders , our customers ,

if you will .

This segmentation strategy was developed by Dr . Chen

to maximize the potential for really getting the knowledge out there

through the organization in a practical and applied way .

What you can see is most of the users within this case study organization ,

over 5 ,000 are tiered to the A minus program .

We wanted to highlight there 's a 20 % growth rate in 2023 .

So we continue to expect growth .

We 're seeing a lot of engagement from the trainee stakeholders .

The A program really focuses on quality engineers ,

quality and reliability . That 's over 300 users .

Then really the cornerstone of this program ,

the Master Black Belts and the Black Belts , highly trained ,

highly educated , many PhDs , probably more than 50 people .

These people are really being funneled through the A plus program ,

which is really a mentor and an instructor program .

These people are really critical for the success of this program because

they 're the local site champions and they 're not just given knowledge ,

but they 're given the tools to be able to apply knowledge with JMP as the tool .

This is , in my mind ,

probably the most important slide of this presentation

because it really focuses on the strategy and vision of this program ,

which just really funnels through all t he details .

From the top level , this goes through all the details of the program .

I want to emphasize these four called key visions .

Cross -functional leadership and vision team building

through a process that actually was codified in ASQ

and other Six Sigma literature,

forming, storming , norming , and performing,

and being data -driven , truly data -driven ,

not just data -driven in lip service or through maybe basic tools like Excel .

Then having— really, this is probably the most important one—

having a long- term vision .

These are the visions .

The audience can look at these bullet points .

But what I wanted to mention is that many people try to embrace these visions ,

but they really lack their embodiment .

If this will resonate , I think ,

with many viewers , is that oftentimes meetings might be cross -functional ,

but the communication style isn 't necessarily cross -functional .

Many teams , in my experience in industry as well,

jumped to this norming phase of this team building

where they skip the forming and the storming phase .

Really, this tendency to jump to norming is driven by the fact that

everybody assumes that they already know what they need to do,

and many people come from a highly educated PhD background .

Just because we have these highly educated people

doesn 't mean we have a strong team .

Bypassing this forming and storming phase

where people really take the time to work through what their roles

and responsibilities are and understand the expectations,

this shortcut ends up actually giving us really a non -long- term payoff .

In fact , it really hurts teams .

The other thing is I mentioned this briefly , but I 'll say it again,

the notion of being data -driven is often very misunderstood .

Many people think that if they present data in Excel that that 's good enough .

But it 's one thing to present the data in Excel ,

but it 's another thing to actually meaningfully present the data

in Excel or any other tool .

Oftentimes , when people assume they already know what they don 't know ,

then this reflects a miss of important details

that they can use to solve their problem better .

The final thing I want to say , which I may not mention ,

but it 's peppered through the rest of the presentation , is that

it 's fundamental that we recognize the achievements of every individual ,

especially the MBBDV mentor candidates ,

because it 's through this recognition that we cultivate passion in the candidate .

The program no longer becomes an obligation or an assignment

from management , but really an honor and an empowerment mechanism

for these individuals to become more capable themselves ,

investing in them, and also give back more to the company .

This is the roadmap .

There 's a lot here , but the key elements here are ,

again , that we 're grouping internal and external material .

When I say internal material ,

I 'm talking about the material in the blue .

This is well -vetted , well -thought -out material

that comes from the product of years of engineering ,

applied engineering experience within this organization

and within the program champion's knowledge base

from previous organizations ,

which is a Six Sigma Black Belt focus .

Then in the red , we have all the JMP education training materials

which have been acquired through NDA by this test case organization .

You can see how strategically , through these A , B , C , D , E,

and scripting language main modules , we 're looking at disseminating

and developing leaders for this program .

Obviously , for any questions about this , please reach Dr . Chen ,

my co -author presenter here .

I 'm also happy to take questions .

One other thing is that it 's about a 50 -50 split between the internal

and external curriculum , if it 's not obvious from this slide .

I just want to make sure that that 's clear .

Then what I 'll show in subsequent slides is you 'll see this 0 .5 nomenclature

and I 'll explain what that means later .

This is a beautiful slide .

There 's a lovely story here , and it 's really about connecting ideas .

But let me say first that we 're going to basically start down here

at the foundation .

The A0 , A1 , A15 , come up to the A2 ,

which is the nucleus , and I 'm going to mention this again,

and then go up to A3 and A4 ,

and then go to B1 to B4 , and then down to C1 to C4 .

It starts on the lower left ,

and then it goes around , counterclockwise, down, and ends back at the lower left .

What are we trying to communicate here ?

The focus , really , as the title of this talk implies ,

is on applied engineering statistics .

There 's five stages . There 's foundation , A0 , A1 , and A1 .5 .

The 0 .5 was identified as a bridge to help bridge the knowledge gap

between the A1 and the A2 curriculum , which we identified .

I 'll just say that here to make that clear .

The A2 , which really focuses on basic statistics

and modeling ANOVA and regression .

That 's up here .

Then connecting the dots , we 're going to the DMAIC curriculum ,

B1 to B4 , which really focus on the DOE material ,

and then from there, progressing on to the data mining

and text mining material ,

which is really what the course content entails .

The progression here , there 's an analogy here drawn

between this training curriculum and kung fu or Chinese kung fu.

Let 's talk about that .

Kung fu really embodies this idea that excellence and mastery in any endeavor

require persistent effort , dedication , and time .

It 's not a quick fix .

There's no you learn everything and you 're an expert .

That 's the idea here .

The emphasis here is that the foundation must be very solid ,

so skill achieved through hard work .

We go through the A0 , A1 , the foundational study ,

we go through the 1 .5 to help bridge up to A2 .

By doing the A0 , A1 and A 1 .5 , we 're going diverse .

There are 72 skills in Chinese kung fu that have to be learned .

I think this illustrates that .

We go diverse in order to build a foundation ,

and then we strengthen the foundation through the A2 ,

through the ANOVA and regression modeling techniques .

Then from here , we get to the A3 and A4 ,

where we 're talking about the central limit theorem

and rational subgrouping , which are really fundamental ,

not just in statistical thinking , but in applied engineering thinking ,

because data is part of engineering now .

What we 're showing here is that

there 's basically two key maybe Chakras or acupuncture points , if you will ,

that require opening up in this martial arts tradition .

Those two points are drawn parallel

to central limit theorem and rational subgrouping .

What I will say is that these concepts ,

while theoretically many people understand practically , they may not .

The central limit theorem and rational subgrouping are actually

closely connected to a process engineers use all the time ,

the CPK and PPK and understanding that is a great launching point .

I think if anyone has any questions about the importance of those ,

we 'd be happy to discuss that .

Rational subgrouping is so important

because it really refers to how we 're going to understand within

versus between variation and how we 're going to meaningfully

sample our data so that we can extract practical insights from it .

So many times in an industry ,

I 've seen examples where sampling is done non -systematically

and in a way that produces meaningless information .

So rational sampling , rational subgrouping is really critical .

All this , up to this point , [inaudible 00:17:21]

the product of developing a very strong foundation ,

connecting the dots ,

and only then do we really move over

to the more advanced DOE and regression tools ,

which is the B1 to B4 program ,

and not just learning the tools , but becoming effective and fast

and using the systematic DMAIC framework to drive them .

After the stakeholder works through the fruit of this endeavor ,

through project work and coursework ,

they can become a master and they can start employing more advanced tools .

Really , the analysis effort becomes more of a creative integration

of different tools and techniques

and with an understanding of their limitations

to be able to solve a problem holistically .

There 's a lot there , but I really wanted to summarize the power there .

Now , here this slide , we 're talking really about

the high -level deployment of the Black Belt training curriculum .

You can see that most of people , again ,

like as shown in the previous slide , are trained in the A0 , A1 , and A2

for the foundation over 800 , over 600 , over 300 .

Then you 've got the MBB, BB mentors who are required to do the project work ,

starting to be trained to be in A3 and A4 .

These programs ultimately need to be driven by local leaders

who promote and empower the folks that are taking these courses under them .

This is the only way that the program becomes truly impactful

in the organization , because the leaders drive the change ,

and through their knowledge and experience ,

they teach people who came before them .

The key thing here is the participation and application in projects is fundamental

for these leaders because only through applied learning

in a project context

can one truly become an effective applied engineering statistician .

This slide , in essence , maps the JMP tools to the DMAIC steps .

For each DMAIC step , we pair it with the JMP platform

so that the stakeholder can consider and learn to solve a problem

in a more systematic manner .

That 's really the power of the DMAIC framework .

Our approach is , I would say , more difficult in the beginning

because the candidate has to get the data first .

That 's difficult at first ,

especially for somebody who isn 't trained in real applied engineering data analysis .

But it ends up being smoother in the end

because once they have the DMAIC tool , they can drive the project to completion .

They 're much more likely to drive it to completion in a systematic manner .

One thing to mention here is that

a lot of people in today 's engineering technology environment

want to get certification in Black Belt .

But oftentimes , especially in high -tech industries ,

the engineering function isn 't well defined .

With a paper certification ,

it 's probably not of any real help to the trainee .

They won 't really learn anything

to be able to actually make an impact in their organization .

The goal of this program is really to build impact , create impact .

In this slide , really the emphasis here is that the A2 class ,

the ANOVA and regression ,

as I talked about as being the fundamental and the bridge ,

it 's important for the global vision , and it really , truly is a global vision .

There 's deployment across the world .

So far in this case study , we 've deployed over 30 A2 instructors

ranging from geographies like Germany , Israel ,

even China , bridging over to Japan .

We 've offered 20 of the A2 in 2023, 20 of the classes .

The first 120 trainees have to do projects .

The remaining trainees , over 180 , have an optional project component ,

but they can 't continue beyond the A2 program as a result of that .

I talked a little bit about the A minus or the 0 .5 .

and so this slide is really to speak to that .

What are the objectives in creating this A minus program ?

It 's really to bridge the gap

between the original A0 and A4 modules that were developed .

The project component is they don 't need to do .

Of course , as I mentioned , not having the data initially makes it difficult

for the training to get started .

This covers that limitation , if you will , for certain people

and addresses that knowledge gap that we identified .

Once they finish this training ,

they can go more deeper into advanced subjects .

Many people come to learn JMP

and they like DMAIC as a problem -solving methodology .

This is a good mechanism to catch them and find them where they are .

Actually , this program is extremely popular right now .

It 's well overbooked .

The plan was to book only 12 people in three of the 1 .5 trainings , for example

and there 's already over 23 or over 25 people .

Again , the local leader , the local MBB leader is facilitating

and maintaining and following up and continuing to generate interest .

That 's actually really fundamental to this program

and the success of this program .

Also , there 's some feedback from people

who already took the more advanced program ,

who are now taking the 1 .5 , and they like it very much .

I think many of them find that it helps them connect the dots more

with what they 've already journeyed through .

This slide just gives you a feel for the contents of this program .

You can see that obviously there 's a Getting Started component .

Graphical analysis and outliers, box plot analysis

is really fundamental if done properly .

Then the A2 , A3 , the ANOVA, regression ,

the MSA measurement systems analysis, and Gauge R&R .

Then , in the A4 , we introduce SPC and Multivariate SPC .

Then in the C program , we get into the advanced data mining ,

text mining , PCA , multivariate methods , all through the JMP 's platforms .

Actually , this type of a program , the goal would ultimately be

aspirationally to develop it at the supplier level

so we can create a synergy with suppliers

and improve quality using this true data -driven approach with JMP .

How would we deliver A minus training classes ?

Well , this slide gives you a high -level overview of how this happens .

This reflects the training style .

We deliver the subject in five steps .

The example isn 't critical in this context but we use Choice Design .

In this case , we use Choice Design to conduct an Attribute GRR

which was also featured later .

We go through the launch window

and demonstrate how to populate it in Choice Design in the DOE .

We conduct the analysis and look at the top -level statistics here ,

the marginal probability and utility .

We interpret the analysis

by looking at these statistics and the probability profiler .

Then based on that , we take appropriate improvement actions .

This is also more about interpretation

then we take appropriate improvement actions .

When the trainees go through this process ,

it actually really helps them identify gaps in their learning and fill them in .

This is where the applied style is really important ,

even in the non -project required curriculum .

This is an expansion of this A -minus program

and emphasizes a really thoughtful and strategic vision where the yellow…

These are actually modules as summarized in previous slides

in a similar manner .

The yellow are the modules that really we 're in the process of developing .

The green ones have already been developed by the program champion ,

mostly , I think , or exclusively by Dr . Chen .

If you jump to the E , it 's really about reliability and marketing .

I mentioned B is the DOE , and the custom DOE really is the emphasis .

C is the abbreviated data mining , and the A1 , A2 are really the foundations .

The emphasis here is that after each MBB gets certified ,

it 's really not the end , it 's really the beginning for them .

There 's an emphasis and careful thought in customizing each of their functions ,

developing them .

Some , for example , may go into DOE . Some might take data mining .

Some will become a leader in the A1 .5 curriculum and so on .

The goal is for us to make sure that all these leaders

ultimately become not only certified , but capable of training

and developing other local leaders .

In the paper certification anyway by the Global VP of Quality and the CEO

is a testament that we really follow these people

in their development path after certification .

If it 's not obvious already , their strength is JMP here

and using JMP to solve their problems .

This is a fun slide .

This is me and my daughter , just over a year old .

But what we 're showing here is that

probably a record number of internal people in this organization

took the STIPS exam at the same time .

There was 100 % passing rate .

These are a median score of 915 , an average of 899 .

Top two scores for these two individuals , very close to 100 %.

There 's , of course , me and my daughter .

Dr . Chen is here and here 's his son .

We have this little JMP girl and JMP boy thing going .

It 's fun .

I 'll credit and thank Sarah Springer

for her really wonderful collaboration at the end of the presentation .

But here she is here .

A key component of this program , just like today ,

is participating in JMP Discovery Summit .

This highlights our Discovery Summit achievements .

The fruits of real work

that are being recognized by Discovery Committee members .

I think this is from a June 12th Gloucester Forum event .

The two key event ingredients I mentioned

are the forum events and the Discovery Summit participation .

This is the June 12th Gloucester JMP event here .

We have asked JMP to support JMP 17 new features .

We 'll continue to seek their support

during an upcoming September 11 Hillsborough event .

We 're engaging with basically all of the JMP Discovery Summit events .

US , Japan , China , Europe .

Let me see .

I forgot to mention, this is Agatha Debris . This is Sarah .

This is one of our candidates , or actually leader at this point .

This is Don McCormick , of course , of JMP .

We anticipate quite a bit of engagement in the 2024 Europe .

Actually , there are many reports that are available .

Half of the 80 that have been developed are confidential ,

so those will probably be off the table .

But we 're definitely expecting some good likely acceptance

based on the work that 's been done .

Here , we see the Singapore event ,

the 2022 September 9 Singapore Elite Eight Tournament event .

This just highlights our competition .

It was a very competitive and successful event .

Talks were very diverse .

I actually presented a talk here .

You 'll see me in here on box plot statistics ,

which was high -quality enough

to be accepted at US Discovery Summit last year .

I 'll probably link that on the Discovery page for this project .

But a key aspect of having these forum events is soliciting feedback

on the presentation quality .

That 's where JMP 's participation comes in the organization .

We seek engagement from JMP stakeholders who can review the material

because it garners enthusiasm and engagement for those presentations .

It makes them more concise , more effectively delivered ,

and it moves to a feedback -forward model

where continuous improvement is obtained through continuous feedback .

There 's this mindset of technology facing , moving to service facing .

This collaboration with JMP and this partnership with JMP

helping us to review these presentations as part of this effort ,

not only improves the presentation quality for the purpose of submitting at Discovery

but it improves the quality of the work

that 's being done at the organizational level .

I 'm going to briefly cover some case studies

just to give you a flavor of what we 're doing on the ground .

This case study was really focused

on comparing Excel versus Minitab versus JMP Gauge R&R analysis .

It 's about how do we manage a destructive Gauge R&R ?

First of all , how do you know a test is destructive ?

On the left here , we basically did ...

In the measure phase , we did a rigorous comparison

and showed that JMP is more reliable for the decision -making process .

The key is because it considers the ANOVA , the analysis of variance

with an interaction ,

and that 's really the most comprehensive and best tool out there .

On the right -hand side , we 're talking about that destructive test methodology

and how do we determine that .

There 's really three approaches .

We can assume that the study is fully crossed with no degradation .

We can assume ...

When I say degradation ,

I mean the sample doesn 't change as it 's being tested repeatedly .

We can assume the test is nested , so there 's some degradation behavior .

We have a third choice .

We can use the crossed methodology .

But if we can systematically go through a decision -making process with a flowchart

to show that the destructive quality on the sample is minimal

within some prescribed limits ,

then we can use this third choice ,

and this flowchart helps manage that decision -making process

to decide how we want to approach the destructive method .

I forgot to highlight .

This presenter became BB -certified and scored 925 in the STIPS exam .

This project owner .

The second case study achieved third place

in the annual rankings within the organization .

The emphasis here is on conducting ,

which I alluded to before conducting an Attribute Gauge R&R

using a Choice Design ,

which is really , in my mind , a very novel application for Choice Design .

It 's very exciting .

The objective overall on the response

was to reduce Failure Analysis cycle time reduction .

This presenter presented in 2020 at the US Discovery Summit

and scored an impressive 925 on their STIPS exam .

Just to give you a sense here ,

the question we 're trying to answer is

how do we know that our team

can make a consensus decision in our meetings in general ?

Considering this Attribute GRR in our model for that

is what 's been done here .

Briefly , if all the members achieve a response probability of 100 %,

then that would imply that the team can make a consensus decision ,

as you can see these numbers here , these response probability numbers .

But what this is showing is that a few of the respondents

that scored higher in the green and at the end , actually , too ,

scored lower .

These respondents who scored higher really dominated the meeting .

They were the most talkative , the most experienced .

These lower -scoring respondents are the people that

either they weren 't paying attention or they weren 't talking intentionally ,

they weren 't engaging in the meeting .

This is a data -driven way to demonstrate

this team is not ready to make a consensus decision right now .

This is another case study .

A lovely case study where we use text mining to search keywords .

Actually , you 'll see ,

part number is part of the word cloud that we identified in Text Explorer .

The novel thing here is we saved the indicators for these words .

By doing this indicator saving , we put part number into the model .

When we put part number into the model ,

that model went from a poor model to actually a very good model .

Part number became the strongest ,

the most effective factor in driving the response here ,

failure analysis cycle time response .

It 's a very powerful and elegant application

of using Text Explorer bread -and -butter to go into modeling

without JMP pro actually .

What 's great about it is it 's very easy to understand

how to work through that workflow in JMP .

One thing that 's super powerful about this

is when we have a model and it predicts well ,

we don 't have to do a bunch of other root cause analysis

that we did up to this point

for other , say , similar part numbers in the product family

where the part number performance would vary

even within that part number product family .

This is another quite exhaustive case study .

The key here is this presenter did scored very , very well

among the top in the STIPS .

They passed the DOE certification exam

and they also presented at Discovery Summit in 2022

and were BB certified .

But they were basically measuring a very difficult shape .

They had to modify their recipe in order to perform that measurement .

This analysis , it 's quite small , but in effect , what it ended up showing

is that they were able to achieve a significant performance

in their precision- to- tolerance ratio ,

so there 's a P- to- TV ratio , and P- to- T ratio .

This shape is very complex .

This is a hallmark in high tech .

Another example here shows

that there 's non -uniformity in dimension on the basis of a location .

We have to perhaps look at the distribution

of the average difference from location to location ,

from center to edge , rather than just trying to take

an overall aggregate measurement to capture the entire surface .

That 's what some of this analysis highlights .

Here 's another really lovely case study .

This one used group orthogonal super-saturated design

to block the first failure mode from the second failure mode

in this problem .

It was a two -step process optimization using this GO -SSD .

The project owner had to figure out

how to manage seven variables in this context .

It took her over two weeks to figure that out .

Basically what we 're showing

is the design process through different [inaudible 00:42:20] she might have considered.

Then on the right, we 're showing some Monte Carlo simulation in JMP ,

which JMP is very user- friendly at doing .

Then essentially validating the optimal process settings

where if we looked at the simulation

were the means within the confidence interval

and the prediction interval range of the model ,

and that was the tool for validation ,

graphical visual interactive tool for validation .

Then the presenter also went into some SVC to identify any process issues .

The key message here is that if the design is good ,

doesn 't mean the process is stable .

You can see this process is drifting .

That 's why the SVC component

is so critical in the trainees' learning path .

Here 's another really lovely case study here .

The emphasis is using SIPOC

with process improvement flowcharts to define the scope of the project .

The interesting thing about this project

is the project scope really pertained to the process itself ,

so we 're in effect modeling a process .

The approach was creating a model and then refining it

and working on cycle time reduction

and improving the model 's predictive capabilities for cycle time .

This emphasizes the importance of developing a robust model ,

so that process automation predictions can be robust .

On the right here , we 're looking at leverage plots .

This is quite interesting

because we see data points that are off the residual by predicted plot here .

I drew some lines in here .

They actually reveal a collinearity issue .

This pattern reveals a collinearity issue which is consistent with the high BIS

that we see in the parameter estimates table .

This reveals a combination of potential things :

hardware constraint , phase constraint ,

a data acquisition problem through the absence of a cyclic pattern,

and that we can see in some time series augmented on a studentized residual plot .

There 's the opportunity here to go into a more detailed time series analysis

to understand why is there this time shift problem ?

Why is there this collinearity problem in time

or in space , for example , for angular depth or wafer location ?

Autocorrelation is fundamental and this project highlights that ,

whether it be multicollinearity in space or autocorrelation in time .

This slide just highlights yet another wonderful forum event ,

2023 March Madness .

It was extremely close and very competitive

and the top eight people were invited to the Singapore forum event .

Here is one more case study .

In this case study, we 're using item analysis

to focus on difficulty and discrimination ,

to basically strategically assign exam questions ,

JMP STIPS questions to either a Green Belt or a Black Belt exam .

By carefully looking at discrimination and difficulty in the context

of this item analysis framework ,

we can effectively categorize the questions .

You can see on the right how we did that .

We selected four questions that showcased

our rigorous selection criteria .

Actually, I 'll show that in the next slide .

But here what you can see

is that discrimination can be seen by the steepness of the curve .

If you see this step ,

you can say , okay , this is a very discriminatory question .

Then difficulty refers to the translation of the curve

and these colors indicate different categories

where we used…

Actually, this is very important , we use this flowchart .

We can dichotomize difficulty as easy and hard

and then discrimination in terms of the location .

I mentioned that .

Then we can dichotomize discrimination

as yes /no in terms of the location and steepness .

The colors show the different categories

and this goes into actually specific examples

of which questions fall into each category .

We can even go a step further

by looking at the patterns within each of the assigned categories .

Actually , you can think of this as group one .

The group one questions were easy

with no discrimination where basically everybody got it right .

The group two questions , easy with discrimination ,

basically, most people got it right .

The group three , hard with discrimination , basically, everybody got them wrong .

Then the four hard with discrimination ,

most people got it wrong , basically .

You can really think about the power of this methodology

for segmenting questions and classifying them

and giving a meaningful and challenging exam to students

to develop them in the right way .

Then these cell plots added

an additional layer of analytic immediacy in JMP where we can see the proportion

of respondents that answer the question correct versus incorrect .

It 's a very nice companion to these item analysis curves .

For reference , blue is incorrect , red is correct .

It 's a little flipped from you might think .

Just a couple more quick case studies and I 'll wrap up .

This case study ,

we used ANOVA and regression to improve supplier quality control .

This is one of our most prolific student's mentors .

Now he got a good DOE cert exam score ,

was Black Belt certified over 900 on the STIPS exam .

Basically , I think the novelty in this project

is really using a regression algorithm— we 'll just jump over to the right—

to handle outlier problems

in the sense of the regression itself or outlier problems in general .

There 's a lot of thinking in here

with respect to different types of regression

and how those different regression methodologies help us

capture outliers on the low and the high end .

I think one thing that 's really important

to think about as a practitioner is ,

is the low end more important or is the high end and why ?

From an engineering sense ,

in this case , the low end was most important .

The fitting approach had to be tailored to the low end .

We 're talking about understanding types of outliers

and not only types of outliers ,

but where they come from , why they originate .

Just a couple more here .

This one is quite nice because it 's a VSM -focused project .

It 's very unique in the implementation of the VSM approach

because mostly people who do VSM ,

it 's a documentation exercise .

It 's quite qualitative .

But in this case , this candidate

who 's also BB certified and taking the BB certification ,

they really went into all sorts of detailed calculations and mapping .

As a result of that , they actually

determined that they didn 't even need to do a Gage R&R

because the VSM was done .

The entire quality metric in this case was VSM- based .

This is really showcased here

because this was the most successful Lean Six Sigma project to date .

Later on this project is actually , I think , still being extended to use SBC

to validate process stability after building the process up like this .

One final case study ,

I think I just want to highlight here that more in the flavor of SPC

using Control Chart Builder ,

JMP 's super powerful and flexible control charting platform

to verify process both stability and capability.

You can see from here

that there 's a certain risk of the process being either not capable or stable enough .

You can see this process floating on the upper spec limit .

The idea is actually , I believe , to collect more data

to characterize the process better ,

to define the spec ranges better , to match the process behavior .

We 're practically looking at that on this chart ,

in addition to looking at the control limits

that are driven by the SVC methodology .

Here , this just showcases , this was the stakeholder , the trainee , their slide .

These folks are really promoting JMP at the local level

and they 're really thinking creatively

about how to promote JMP more themselves , which is fantastic .

Here , this is a coming mid -autumn…

Dr . Chen is calling it a mid -autumn festival .

This is upcoming here .

You can just see there 's just a lot of great thought going into what 's coming ,

who 's involved ,

and it 's really a group effort with these top level MBBs

working with the program champion to drive this initiative .

This slide shows an emphasis on really to become a top five performer

in this healthy competition framework , it 's really not easy .

All these people are really close right now ,

but the requirements are very multidisciplinary .

There 's a participation in the forum events ,

there 's participation in instruction and mentorship ,

in attending Discovery Summit conferences, and preparing curriculum,

and getting the certification .

Even promoting the internal initiatives

through writing articles and distributing them

through the organization .

Of course , assisting in forum events and even finding additional trainees

in the vein of the marketing methodology

and training material that will be coming as part of this program itself .

This healthy competition really demonstrates

the program 's reach and impact

because these people are really working not for themselves ,

but for the entire organization .

Realize I 've gone over a little bit ,

but I 'll wrap up very quickly , very soon.

I think I just wanted to emphasize

that this initiative is getting internal recognition company- wide .

The takeaway here is that the internal education system is now going to host

a lot of the material that 's been developed here .

So the material won 't be free anymore

as it currently is technically in the cost structure of the organization .

The money that 's charged

will support instructors for their trips and their training .

The money collected by this internal education system framework

will be controlled by the program champion ,

and that program champion will assign it to instructors

based on their participation and involvement .

So there can be a transfer of money through corporate cost center entities .

I think many people know that when there 's money lined up formally

within the organizational structure , people really take it seriously .

I 'm very excited about this .

I know Dr . Chen is super excited about this

because there 's a potential for a huge amount of money

inside the organization behind this ,

and that means an increasing amount of influence .

Also, I wanted to say that JMP accounts ,

Sarah Springer has been just instrumental in working with Charles on this

and in making sure that this whole process is working effectively

when considering the JMP component of the curriculum .

I 'm just about done and I just want to present this last slide

and do a little bit of a quick acknowledgment .

The key here is moving from this antiquated Black Belt program

that other organizations have used to this really JMP -centered ,

multidisciplinary applied engineering statistics program

that really emphasizes

gradually empowering leaders purposefully and sincerely ,

generating core values , proliferating those core values

and using them to really drive the program and grow it .

The most exciting thing for me as a JMP employee

and a former industry practitioner for 15 years is synergizing

with industry and synergizing with JMP to maximize the impact

of both the JMP internal materials and the company internal materials .

With that , I think I 've mentioned Sarah many times .

We really appreciate what she 's done .

Peter Hersch and Don McCormack have been great

in terms of deploying trainings for the new features on JMP 17 .

Then there 's a number of people

from the different sites that we 've just called out by name here , JMP Europe ,

Agatha has been very instrumental .

Of course the case study presenters who we 've referred to anonymously here ,

but hopefully, you can see that they 've done some amazing work .

Here 's the JMP girl and JMP boy again .

My lovely daughter and Charles 's son Mason here

several months ago now .

With that , thank you so much for your time .

It 's been a pleasure .

Any questions , please reach out to us .

Thank you .

Advanced Statistical ModelingBasic Data Analysis and ModelingConsumer and Market ResearchContent OrganizationData AccessData Blending and CleanupDesign of ExperimentsPredictive Modeling and Machine LearningQuality and Process EngineeringReliability Analysis