#156 Becoming a Data Ready Organization

Subscribe to get the latest

on Mon Sep 04 2023 17:00:00 GMT-0700 (Pacific Daylight Time)

with Ron Fritzemeier, Darren W Pulsipher,

In the podcast episode, retired Rear Admiral Ron Fritzmeier joins host Darren Pulsipher to discuss the importance of data management in the context of generative artificial intelligence (AI). With a background in electrical engineering and extensive experience in the cyber and cybersecurity fields, Ron provides valuable insights into the evolving field of data management and its critical role in organizational success in the digital age.

Keywords

#datamanagement #automation #dataquality #strategicanalytics #generativeai #digitaltransformation #datadriveninsights #datareadiness #innovation #decisionmaking #technologytrends #businessintelligence #datastrategy #analytics #bigdata #continuouslearning #operationalefficiency #dataoptimization #datainnovation #emrbacingdigital #edt156

Listen Here

Evolution of Data Management: From Manual to Automation

Ron begins the conversation by highlighting the manual and labor-intensive process of data management in the early days of his career. In industries like nuclear weapons systems and space, data management required meticulous manual effort due to the high reliability and complexity of systems. However, as the world has become more data-driven and reliant on technology, organizations have recognized the need to transform data into more usable and effective ways.

Challenges in Data Management: Complexity and Quality

Ron shares a compelling example from his experience in the Navy, discussing the challenges of managing data for ships during maintenance and modernization cycles. The complexity of ship systems and the harsh maritime environment make thorough data analysis and planning crucial for successful maintenance and repairs. This highlights the importance of data quality and its impact on operational efficiency and decision-making.

Data Readiness and Automation

Taking advantage of automation requires organizations to focus on data quality. In the automated analysis and assessment process, any errors or missing data become critical. To address this, organizations need to improve data collection from the start. By designing systems that make data collection easier and considering the person collecting the data as a customer, organizations can minimize errors and improve data quality.

A holistic approach to data readiness is also crucial. This involves recognizing the different stages of data readiness, from collection to management and processing. By continually improving in each area, organizations can ensure that their data is of high quality and ready to support various operations and technologies like generative AI.

Filtering the Noise: Strategic Data Analytics

Data analytics plays a vital role in driving strategic value for organizations. Ron and Darren discuss the importance of filtering data based on relevance to objectives and focusing on what is truly important. Not all data will be valuable or necessary for analysis, and organizations should align their data collection with their goals to avoid wasting resources.

Furthermore, the conversation emphasizes that data doesn’t have to be perfect to be useful. While precision and accuracy are important in some cases, “good enough” data can still provide valuable insights. By recognizing the value of a range of data, organizations can avoid striving for unattainable perfection and focus on leveraging the insights available.

Uncovering Unexpected Value: Embracing Possibilities

The podcast also explores the potential of generative AI in enhancing data collection. By using interactive forms and conversational interfaces, organizations can gather more meaningful information and uncover new insights. This opens up possibilities for improved data analysis and decision-making, particularly in areas where data collection is crucial.

The discussion concludes with the reminder that data analytics is a journey of continuous learning. Organizations should be open to exploring new technologies and approaches, always seeking to discover unexpected value in their data.

Conclusion

In an increasingly data-driven world, becoming a data-ready organization is crucial for success. By understanding the evolution of data management, focusing on data quality and readiness, and embracing the possibilities of strategic data analytics, organizations can unlock the power of data to drive innovation, optimize operations, and make informed decisions. This podcast episode provides valuable insights and highlights the importance of data management and analytics in the digital age.

Podcast Transcript

Hello, this is Darren

Pulsipher, chief solution,architect of public sector at Intel.

And welcome to Embracing

Digital Transformation,where we investigate effective change,leveragingpeople process and technology.

On today's episode,becoming a data ready organizationwith special guest retired Rear

Admiral Ron

Fritzmeier.

Ron, welcome back to the show.

Thanks for having me.

Glad to be here.

Hey, Ron,it's always a pleasure talking to you.

You come from a great heritage andand on the intel teamwith lots of other former Navy.

Give us a little bit of your backgroundso our audience can understand morewho they're listening to today. Sure.

By Angie.

I am an electrical engineer,and I actually started out my careerin microelectronicsfor the government Sandia Labs.

And while I was at the labs, I actuallyfound out about the Navy.

And I won't go into all the details of whythat was of interest to me, but I joinedthe Navy actually as a reserve.

So I had for much of my careerkind of a dual governmentcivilian and then maybe career.

Turned out that by chance,when the Navy decidedto promote me to Rear Admiral.

They also said, Oh, by the way,we want you full time.

And so I actually finished my careerfor several years on active duty.

Kind of reverses the normal waythat works.

I see.

Admiral Fitzmorris, you did it backwards.

I did it backwards.

But it worked out well for my wifebecause my wife

I married it just before

I went on duty as a fire officer.

And so she didn't know anything aboutthe military except being flag officers.

Why she stepped up to the taskbecause it was pretty amazing, actually,how she jumped.

That's that's pretty incredible.

So a background in electrical engineering.

But you're you're kind of a cyber guy.

Yes. So early on in my careerbecause of really the microelectronicsthat turned out, one of mysuper duper skills,if you will, was I was great advantage.

And that that got me intoreverse engineeringand sort of the black hat businesswhich I won't go into theall the details there but it was basicallyhow things can go wrongwhen you're intendingto try and make them secure and operable.

And that just seemed like a natural piecefalling into sort of the cyber world.

So in fairly short order, I found myselfdoing more instead of all the cyberspace.

There are other things,although I did a lot of comms and other C4type engineering work, of course,but the heavy emphasis on cyber and cybersecurity, in fact,even in my civilian career,that ended up beinga large part of what I diddriving operations where we werefrankly taking events that other peoplearen't good at their cybersecurity,as well as trying to improvenational security

Agency were not NSA, butprivate speaking national securityagencies,cyber security,so that that led into my Navy portionof my career where I was the chiefengineer out of the Navy Systems Commandthat today is known as that warand also spent time at STRATCOM helpingdrive that in C three enterprisemodernization.

So great, great pedigree.

Obviously, I'mtalking to someone who knows their stuff,but we're not talking cybersecurity

A today.

We're talking the big, huge push right nowthat we're seeing everywhere is

I specifically generative AI.

But to make that work,we need data behind it.

And managing datais a is a keyaspect to that, both on the cyber sideas well as on operationsand especially now with generative AI.

What's yourwhat's your take on on data management?

Is is it a well known art?

Why I mean where weat with the state of the art?

So even fromthe early days of my careerwhen we were had to basically certifycompanies that were going into this casenuclear weaponssystems or spaces, so you have thesereally high reliability requirements.

There was always this notionthat you had to maintain, developand maintain all kinds of dataas part of validationand certification packagesof whatever those systems or sources were.

But all of it looking back onvery, very manual intensive life.

And so I had experienceskind of almost in crisisthroughout my careerwhere I saw this notion of as we movedmore and more into sort of a data drivenand I'll say i.t driven worldor information technologysort of underpinning not just nice to havebut is fundamentally what enables youto drive whatever your business.

In my case it was all thesesort of national security systemsis that we started with this notion ofhow do Iturn my data into somethingthat I use more and more effectivelyand sort of this

IP automated data world in generalfor example, one for the Navywould be how we would go aboutpreparing for shipswhen they're coming back from deploymentand they're going to come into a maintenance and modernization cycle.

And the issue isthat while that's true, is and how.

How often does yeah, how often does thatmaintenance cycle happen for these ships?

Because a lot of people don't.

And it dependsthere's actually multiple cycle.

So after every deployment there isby definition some level of maintenanceor maybe very, very minimal is a ship mayin fact I'll say come homeand turn right around in a matter of daysor a few weeks.

But then there's like major modernizationcycles and sort of a functionof the different ship systemsand so on and expectations how long thingswill operatebefore they break down to the pointwhere you have to replace or do majorrepairs.

Right?

So it's a variation of those cycles,but I'm talking about cyclesthat typically would happen in the

Worship has been outmaybe a few deploymentsor they've had minor work donein between on those deployments,but now they're coming back forsome kind of what would be considered a aa major modernization or major repair,what we would call on abe a major ship availability.

And in order to preparefor that availability, as you can imagine,these areour ships are very, very complex systems.

I mean, some of them are, for all intentsand purposes, small floating cities.

Right?

So imagine all of the systemsthat are involved in operating that ship.

And oh, by the way,in case you're not aware of it,the maritime environment is a very,very tough environment.

It's hard on everything. Right.

So when ships come back,there's a lot of engineering and planningthat goes into those missiles.

But the first thing they need iswhat is the actual state of the shiptoday? Right.

It's you need to know whathow it was planned,but you also need to knowwhat's happened to the ship over time.

And as you might imagine,nothing stays terribly static on a shipthroughout a deployment. Right. Isyou got to keep it rocks.

All kinds of things happen.

So part of the availability of preparationis you actually send outa team of people who essentially go outand do all kinds of measurementsand preparation for modernization,where you're going to be takingoff systems,putting a new stuff eventually,but also assessing the current stateof all the systems on the shipas best you can,because it's still underway, right?

So you can do everythingyou can really everything apartto the extentthat you can assess the state of thingsthat allows for the plannersback home to be preparing packagesfor it to actually be done.

The availability the issue becomesthat has historically been it'sactually largegroups of people that are flowing outto the ship along with ships force.

Right.

I tend to do all these measurementsassessmentsand you can imagine these guys basicallywith tape measures and put boards and.

Right. It's not right.

I mean, it's it's that hard core.

But every one of those thingsbrings in the opportunityfrom a human standpoint for human error.

And so this course,

I mean, what we've seen is that youexpand, there's going to be errorsand missing dataand all those kinds of things.

And so how do you deal with thatwhen you're just dealingwith the data manually?

You sort of expect that. And that'sjust a natural part of the process.

But as you try and automate analysisand assessment to move things forward,the quality of yourdata becomes even more importantas you look to try and build processes,take advantage of it,move yourself forward.

So that's an interesting point.

I want to

I want to emphasize a little bit.

Your data collection is criticalfor taking advantage of automationbecause in the past,before you would look at automation,there were people involvedso they they could make up the differencein the lack of data collected, right?

Because, oh,

I know something about this ship already.

I, you know, they measuredthey measured something that I knowis not 100 yards, it's only 90 yards,you know, things like that,because there's some tribal knowledgeor some tacit knowledgethat's sitting there.

So but when you automate, like you said,all of a suddenthat those nuances in the dataare completely lost.

Right.

And the gaps in the data are now causinggaps in your analysis as well.

And another thing that I saw overtime was even in placeswhere we had startedbuilding systems to better collect data,there was the notion of there'sstill a human in the loopfor the data collection processat some level.

And so we had systemswhere we started saying,okay, now that we've we've actually builta fairly substantial data collectionsystem, let's talk about how we analyzethe data differently and move forwardmore automation to speed up the processand to frankly,take better advantage of the data.

That was once again, where we learned thathow you go about collecting dataactually makes a big difference.

And one, for instance, iswhen you have a situationwhere an I'll use it, for example,you have a maintainerthat's looking at a platform.

I'm actually

I was actually an aircraft platform,but it's the same across the board, right?

And I've been given this formand they have to fill out.

Right.

And then they input the formon a computer.

So now we have computerizeddata collection, but the guy still saysnow the keyboard and type he's typing and,and because as we we've designed it,we ask them different kinds of questionsthat in many casesare subjectiveand sometimes maintenance as well.

I only know the answerthat I have to fill in the fieldbecause we put into the computer programthat collects the dataa requirement, right?

So we end up witha data set sometime of says I love more.

It's like, why?

Let me ask the screen writer.

It let me get past the screen.

And I didn't want to do something.

That is hilarious.

But but you're right. But but,but you're right. Right?

I mean, this totally makes sense.

So what do you do?

Do we want to take the humanout of the loop?

Because humans, we areunpredictable.

We go around things.

If we don't know something,we we just make something up,right?

I mean, were horrible were horrible atcontent generation that's consumer right.

So so to me there are there are a coupleof things that came out of that.

One was and once again,this might sound better working togetherwhen I've seen this happen so many times,we we don't oftentimeswhen we're trying to build these systemsto help people fill out the datathat we need, we don't actually thinkvery much about the person who's actuallydoing that front end collection.

Right?

And so we get the.

Sense. That it's it's almost like

I think of, you know,websites of today people go to withouteven thinking about they go shopping.

Right.

That's the science behindhow a person, county clerks,a person has to produceor make in order to get to church.

Yeah, Yeah.

In so many cases in our systemswhere when we don't think of that personat the front end of our dataas really being a customer,in that sense, we don't necessarily designthe systems to make it easy.

And so my experience has beenif you, if you don't make it easyfor the person to do the right thing,by definition, you made it easierfor them to do the wrong thingin terms of the data collection.

And and thus you have oneof the challenges of being a data readyorganization is you need to thinkabout the entire lifecycle of your datafrom the very start of collectionall the way through how you manage itand process itto take strategic advantage of it.

Right.

And just

I think it's important to recognizethat there are several stagesof all of that data readinessthat you want to work through.

There's probably roomfor most organizationsto improve in every one of those areas.

I like that approachfrom because what typically happensis people focus on one areaindependent of the others, right?

Oh, we'll do better data collection,give all those guyslighter on the shipso they walk in with onrate.

There's still subjective things, right,that they have toeven if you gave visualization to themlater and a cameraand they walked around the shipthere's things that they're going to smellor see or, you know, feelthat camera and other sensorsjust can't do it.

Not yet.

Just give it some time, I guess.

So that's an interesting

I like how you're you'repulling that all together.

What do you do as far as oncethe data is collected?

There'sgoing to be some dirtiness in the data.

Right? Right.

Even in fully automated systemslike cyberthreat detectionis a great example.

There's lots of noise in in inthat data that's generated.

What I mean,what what do you do about all that noise?

How do you filter it up?

Do you keep it all?

This is a big question

I know a lot of people have.

They're afraid to throw awayanything for fear that that datathat they have may have some nuggets ofinsight in there that they're missing out.

Yeah,

I think this is maybe sort of the magic.

And I had a data scientist work for mewhen I was back at Strathmore.

Now, now,as I mentioned, and in his counsel, it'salways really good as if so many peoplewhen they when they sort of get it.

I'll tell you, data, religion, we're nowgoing to be a data driven organization.

We're data. Driven. Yeah. Yeah.

They without thinking about theythey imagine that they're going to haveoil in the ocean

And he said, you know, let's start withget a teaspoon and say,what do I want to do with this teaspoon?

All of this data right?

What am I going to do with it?

And so having that driving with intent ofwhat makes my missionmore successful, eitherimproving efficiencyor improving performance or bothor reducingsome other costs are trying to understandwhat is your objectiveand using the data to strategicallydrive your business and thenand then do that assert almost back.

So what what dataactually helps me do that.

So there's there is,

I think, a real analytic process of sayinghow do I expect to use my datato achieve mission success?

Right?

And if you look really hard of that,

I think what we will find is thatyou may be collecting datathat is actually not helpingyou drive it to those answers.

Right.

Or as youpoint out, you may have some dirty dataand then you need to ask the question.

I think if you're doing itin sort of a focused analysis,we're not trying to boil the ocean.

You could say, okay,if I have a lot of this data,but it's only 80% quality,whatever that would be, right?

Because it still drives mebased on my analysis to a useful answer.

And again, without going into specifics,there was a great maybe example of thiswhere we had a very large corpusof maintenance data on a systemand we started saying, what can we doto see the maintenance processright thereimproves aspects of the maintenance.

And as we startedtrying to look at the data and say,where do we see indicators of events thatif we made a different decision there,it would improve our maintenance cycle.

We saw one, we had dirty data,but two, we had a lot of data.

And so either with,

I'll say, the dirtiness of the data,we were kind of able with analyticsto overcome that.

So you don'tnecessarily have perfect data,but you need to understand,do your analyticsstill seem to drive youto a reasonable and to make.

So and so?

I like that, right?

You talked about finding outwhere your data is, what data you haveaccess to, and how you can leveragethat data to get analytics.

And I love how I love the conceptof my data doesn't have to be perfectbecause I think a lot of times,especially data scientists, will do this.

I've got to cleanse all my databecause it's a science, right?

The science is alleverything is well known.

There are no unknowns, right?

That's how science likes to work.

But there's this data engineering.

As engineers,

I'm I've got a six degree and a doubledegree.

And so it's it's fuzzy.

Right? Right.

And that's somethingmy professors taught me in engineering.

You don't need to know exact

You need to know withina range. Right.

So that's that's what that'swhere we have to get to with data as well.

Otherwise, we could be stucktrying to cleanse our dataand spending all of our timecleaning our data without getting anyany valuable information out.

And so the answer is it depends.

It's just the classic lawyer example.

There are placeswhere the answer has to beincredibly precise and accurate,and it's.

Like targeting systems of nuclear. Right.

And so, well, maybe notdepending on the nation that the natureof the analytics you're using,you may actually have some systemswhere you really do have to do a datapreparation on your data collectionto ensurethat you don't drive to a wrong answer.

But as you point out,there are just so many caseswhere that's just simply probablynot really true, right?

Is that you can actually get it very goodanalytic results out of good enough.

Good enough, good enough, right?

Yeah.

And and then the other thing that I sawin this process waspeople,even though they collect all this data,don't really knowwhat they've done with it.

So so the notion of actually havingthis data library, in other words,there's actually valuein knowing what you've donewith your data from collection.

All the way through analysis.

And I think that's anotherpart of the datareadiness is kind of maturingin how you actually handle your data.

And I guess I look at itthis way is that if the data really doesrepresent strategic valueto your organization,then it's worth paying attention to it,just like you would your money right?

You don't just go,

Oh, I collected a bunch of money.

I earned a bunch of money.

I don't know where I put it,but it's around or somewhere.

Most people don't do that, right?

Unless you honestly knowmost people don't.

Unless you have tons of it. Right.

And maybe then. Maybethat's a problem. Issue.

Yeah.

I mean, we we generate so much data,right, that and we collect it alland we store it.

And because I feel that wayabout my own data, but right.

I've got a terabyte up in up in Azurecloud of personal data, those arefamily videos and picturesand my podcasts are, are stored up thereand I blew through a terabyteand I was starting to run out of space,right?

So all of a sudden

I started caring about data.

I had a whole bunch of clips that were,you know, copies of clips and things.

So I started going through all my dataand removing duplicatesand things like that.

So it's it's funny when you havewhen you have abundance,but you don't value it as much, right?

I so maybe that's part of the problemthough.

On the other hand,when I go,

I know that there was this picturethat was really,really cool that I think, gosh, in twoor three years ago,how could I find it, right?

Oh my goodness.

I do want to mention one other thingthat I thinkand it's almost counterintuitivein some ways to what we what I saidearlier about I think the waythat a lot of organizations can moveforward is by doing really oceanstart focused and let that grow.

But even as you start on small things andand focus on it, don't losesight of the big picture.

And I say that because one of the thingsthat I've seen time and time againwhen we start trying to do real dataanalytics, right,to drive decisionsthat drive strategic value,we also discover value that we weren'teven looking for in the data.

Right.

And and that's kind of like the Azure goalthat you really weren't expecting,but then you just kind of run into.

But it's importantto have that kind of mindset of thatwhile I'm working on a focused set,it was an intended outcome.

I'm always trying to keep an eye on what'swhat'sthe bigger picture of my strategic goalsfor the organizationand pay attentionto that again, back at me, it's example.

One of the things that welearned out of that study besides the factwe have a lot of dirty data,besides the fact that we have so much datathat we actually were ableto get some good analyticsdespite the fact we have some dirty databecause we learned all the early thingsaboutthe collection process to dois just inherently trying to improve that.

But what we also foundwere some interesting things aboutwhat was our hypothesisof what needed to be doneto improve maintenance versuswhat the data started showing.

And one of the thingsthe data showed us wasthere are certain partsyou send back for maintenance and repairthat are just lots,which is to say you're going toin that part and it's going to be repairedand it's going to come backinto the stockpile and you're on it.

It's going to be backsometime really soon.

In other words, there's something abouthow we are a process of repair and tasksthat is certainly not sufficient,but we haven't nailed it down yet.

So you can use that one to say, okay,that means there's somethingwe need to learn abouthow we do that repair and that testing.

But you can also say at some levelit's actually cheaper just to say,you know what, if I had a path that passespast a certain pointof how many times it goes for repair,throw it away,it's actually cheaper and fasterjust to get it out of the stockpilethan it is to keep finding itand let it become a bigger partof your maintenance cycle.

And probably that wasn'tnecessarily those.

Those were great.

Yeah, those are great nuggets, right?

When you when you actually run into thatright wherewow that I learned something newso I like your idea is stay focusedbut keep one eyewandering around and looking athey what other additional thingscan I learn from the data?

Great, great insight in the data,especially now that data is becomingwe've heard data, data'sthe new oil, whatever.

But now with generative AI,

I think it's starting to unleashsome of that data even more.

And like you said before,like I can see some new things happeningwhen someone's filling out a formthat it's not just a static formanymore, that again, I could be askingquestions and converse with,with the person to get information out.

Like if they sayif they say, hey,what's the status of this?

You know, do hickey on on on the aircraft.

And they go, I love my mom.

It's going to go, That's great.

You love your mom.

Can you look at this thing, please?

So is it black?

Is it white? Is it pink?

Is it you know, it'ssomething more interactive toto spur on and help people get over there.

And in placeswhere that data collection really isalso a big deal because in some cases,maybe you really don't care that much.

But there are a number of placeswhere it really matters.

And I've seen systems where we say,well, let's let's reducehow much free text input we ask for.

And so what that, you know, is we now goand we say, okay, now you pick lessand you have to pick one of these things.

And I'll tell you, inevitablywhen I get confronted with a pick list,you know, the majority of the time.

I don't pick the second. One.

No, none of these things in the list this,but I can't get past it.

And so I pick one out.

And so if I had some abilityfor those thingsthat really matter to say, let's interactwith that person at some level and sayso back to me, well, why don't thesewhat's something to help builda level of understanding that that can bereally, really cool in my mind.

Right.

And has seen another great use casefor generative AI to help inin data collection in getting informationfrom from the real world.

Ron, it'salways a pleasure talking to you.

We always have fun talking.

Thanks for coming on the show today.

Thank you,

Darren, I appreciate it. Every day.

Thank you for listening to Embracing

Digital Transformation today.

If you enjoyed our podcast,give it five stars on your favoritepodcasting site or YouTube channel,you can find out more informationabout embracing digital transformationand embracingdigital.org Until nexttime, go out and do something wonderful.