Latest / The Effective Data Scientist / The Secret Power of Data Evangelist

Transcript

The Secret Power of Data Evangelist · The Effective Data Scientist

• 0:07
  Speaker 2: Welcome to the Effective Data
• 0:20
  Speaker 0: Scientist Podcast.
• 0:21
  Speaker 2: This podcast is designed
• 0:50
  Speaker 1: Today.
• 0:51
  Speaker 1: I'm here with Paolo and Jeff.
• 0:54
  Speaker 1: Hello, Paolo!
• 1:21
  Speaker 1: This episode is about the secret power of Data Evangelist and it's honestly, I've been looking forward for a really long time because today we have this episode with my colleague.
• 1:38
  Speaker 1: With my business partner was my mentor in my true friend Jeff Stevenson who isn't who is a statistician like me and working for Burring Engelheim.
• 1:52
  Speaker 1: Hello Jeff, welcome!
• 1:55
  Speaker 0: Hi Aziza thank you.
• 1:56
  Speaker 0: uh Thank You For Having Me on here.
• 1:59
  Speaker 0: just a real quick story though.
• 2:02
  Speaker 0: when I started with BI three years ago i met Aziza one of the first things And I realized we were kindred spirits And I've actually used that title Data Evangelist before, so when i mentioned that Aziza started laughing.
• 2:20
  Speaker 1: Yes it's always great to meet and find colleagues who share same passion and the same energy.
• 2:28
  Speaker 1: this was really a great that you joined us at um.
• 2:32
  Speaker 1: It was very heartwarming Always with you and you changed the vibe of our organization And I believe you are a change maker.
• 2:44
  Speaker 0: Oh, well thank you Ed and what i can say same for you if i'm going to do something i want to make a difference
• 2:53
  Speaker 1: yeah yes and making difference and making impact that will last long and um not only just doing things but doing significant.
• 3:06
  Speaker 1: It's
• 3:11
  Speaker 2: very nice to meet you.
• 3:13
  Speaker 2: I was very curious to record this episode because i'm very curious about this definition which sounds fascinating.
• 3:25
  Speaker 2: and uh,I'd like to know more about you for example About your background Your education What drives you and what are the key, I don't know elements in your individual story that made you a statistician.
• 3:50
  Speaker 0: Okay well i'll try to keep it short because I love to talk but my background is math Baxlers and Masters in Math, and Emphasis and Statistics.
• 4:07
  Speaker 0: And then I went another year toward my PhD in statistics.
• 4:10
  Speaker 0: so... My whole career has been mainly in the world of science and engineering So it's always been with a physics or engineering flavor.
• 4:22
  Speaker 0: There's some chemistry involved but With BI its'a lot biology involved.
• 4:28
  Speaker 0: so i'm having to learn that which defi- one of the things that really defines me is I love to learn.
• 4:35
  Speaker 0: So, i'm also a teacher at heart.
• 4:38
  Speaker 0: uh...I've taught for many years and adjunct faculty because I love too just inspire others as what I like to do and that sort goes along with statistician than I am .
• 4:52
  Speaker 0: I guess one thing it really stands out about I've worked with a lot of very good companies and companies that had one statistician or companies who may have had six or eight in a group.
• 5:08
  Speaker 0: And, i've always known that i've been different because my philosophy is this if to the customer, the engineer or scientist.
• 5:25
  Speaker 0: If I cannot provide value and guidance in analysis because of a normality issue or some academic approach then really what good am i?
• 5:42
  Speaker 0: So if there's assumptions that can't meet...I need be able walk through with scientists and engineers instead of just using that as a hurdle to say, we really can't do anything.
• 5:58
  Speaker 0: So that's what really sets me as unique is that I work with the customer to try and solve problems.
• 6:08
  Speaker 0: That also brings up good point...I've always said The more understand process ...the better value i am as a statistician.
• 6:20
  Speaker 0: It's hard to be a good statistician if all you're doing is looking at numbers.
• 6:24
  Speaker 0: You need to be able to understand not only the process, the manufacturing process but The people and also the measurement processes.
• 6:36
  Speaker 0: really where my focus.
• 6:37
  Speaker 0: Yes Great
• 6:44
  Speaker 1: Yeah I remember Jeff as you mentioned.
• 6:48
  Speaker 1: Yeah, you have this inspiring capability And they remember one of their common or joined initiatives that we were working is working with our job partners to adopt the steps course, which is a statistical thinking and problem solving.
• 7:09
  Speaker 1: industrial problem solving which included statistical thinking, exploratory data analysis quality methods decision making based on beta correlation and integrations.
• 7:20
  Speaker 1: And one of the main topics was design of experiments and predictive modeling in text mining.
• 7:27
  Speaker 1: so it inspiring how you were managing and how you are supporting this the jump community in our organization to really have access to the best sources for learning.
• 7:42
  Speaker 1: And working together, share all knowledge... This was one of successful initiatives I guess which worked well.
• 7:52
  Speaker 1: what do think?
• 7:54
  Speaker 0: Well i appreciate that!
• 7:58
  Speaker 0: I am not the absolute smartest statistician in
• 8:02
  Speaker 1: the world,
• 8:03
  Speaker 0: but i'm very determined.
• 8:06
  Speaker 0: And if don't understand something or there's a hurdle then ask questions because these are real concerns with people and you usually realize that You don't have to be the smartest person.
• 8:27
  Speaker 0: All you have do is, you have to Be respectful of others and understand their process And go ahead ask your questions because there's not a dumb question.
• 8:40
  Speaker 0: So I'm sort that almost where they are an advocate in which i advocate for The scientists engineers who can necessarily get to the data because they may not have programming skills, and this is where I'm an advocate for them.
• 9:03
  Speaker 0: So part of The Data Evangelist that I am also about getting analytic ready data to end users so that they can solve problems faster if every single one of them and I'm talking about very highly complex organizations, everyone have a similar weakness.
• 9:31
  Speaker 0: And that is they have data in disparate sources.
• 9:36
  Speaker 0: systems there are different places that we have in all of the commitment to AI and And all the money they put into R&D.
• 9:57
  Speaker 0: I don't understand why it's not a very high strategic plan To have integrated clean data.
• 10:09
  Speaker 0: A lot of companies do have some clean data, but it just is not integrated.
• 10:13
  Speaker 0: It's not part of this strategy.
• 10:15
  Speaker 0: So whenever you buy new equipment whenever you have a discussion on new analytic methods or measurements, part of that discussions should always be does it connect with our integrated system?
• 10:29
  Speaker 0: And that is one thing I think we share amongst everybody who's listening to this podcast.
• 10:36
  Speaker 0: Is the data is hard to get too and its'e hard clean.
• 10:46
  Speaker 0: That's what makes them have to come to either IT or they come through the statistician.
• 10:51
  Speaker 1: So
• 10:52
  Speaker 0: we just get data for then and get it clean.
• 10:56
  Speaker 0: Seventy percent of my job, the last thirty-five years has been getting data in cleaning up for somebody else And that is not necessarily a fun part but as most value because The statistician is not going to be the one solving problems.
• 11:15
  Speaker 0: It's gonna be the ones closest to the process.
• 11:19
  Speaker 1: Is it
• 11:20
  Speaker 2: true also for clinical trials data and you see space for improvement in that area?
• 11:29
  Speaker 2: I mean, we know there have been many progressives in the field like this level of standardization the level of the software we use, but do you think that there is still some room for improvement in that space?
• 11:55
  Speaker 0: That's a good question Paolo.
• 11:57
  Speaker 0: I'm sure it is!
• 11:59
  Speaker 0: i have absolutely no experience with clinical trials though...that's a great point because as a statistician including the sources of variation, which you're typical.
• 12:16
  Speaker 0: You should call a diagram your five M's whatever you want to call it but I don't have any experience with clinical trials.
• 12:24
  Speaker 0: But i'm sure there is opportunity in clinical trials for better data management.
• 12:31
  Speaker 0: So that was good question.
• 12:37
  Speaker 2: What is activity process statistician?
• 12:40
  Speaker 2: because maybe not everybody is familiar The job.
• 12:48
  Speaker 0: Um, a process statistician.
• 12:50
  Speaker 0: to me what I found my niche in is being uh half engineer have statistician You know sort of like data scientist data scientists as the new thing.
• 13:01
  Speaker 0: that's title.
• 13:03
  Speaker 0: That came out the last ten years.
• 13:05
  Speaker 0: and What does this you?
• 13:06
  Speaker 0: Know you don't see the term statistician much anymore But it data scientist.
• 13:10
  Speaker 0: when i heard a data scientist was It Was A statistician who does just a little bit more programming than the normal statistician, or it was a programmer did.
• 13:21
  Speaker 0: Just A Little Bit More Statistics Than The Normal Programmer.
• 13:25
  Speaker 0: Okay well me I am really more of a statistical engineer.
• 13:29
  Speaker 0: i deal with the manufacturing processes and everything to be is about variation because that's what my passion getting data clean, reducing variation.
• 13:50
  Speaker 0: And the one place that I really have zoomed in on is The Measurement System Variation.
• 13:56
  Speaker 0: Some industries call it gauge R&R which is repeatability and reproducibility.
• 14:01
  Speaker 0: others called measurement system analysis.
• 14:05
  Speaker 0: but everything to me comes down To measurements systems.
• 14:09
  Speaker 0: That's the first place.
• 14:11
  Speaker 0: always look because especially bio like I'm in now, you have so much variation in your measurements.
• 14:19
  Speaker 0: If you can't trust your measurement how are you going to be able to tell if there's a process change upstream?
• 14:24
  Speaker 0: You know when your-your measurement is so varying?
• 14:27
  Speaker 0: So There're two parts the gauge R&R.
• 14:30
  Speaker 0: There's repeatability and reproducibility.
• 14:34
  Speaker 0: The very first place for me to start Is the simplest And i call it a sanity check.
• 14:39
  Speaker 0: It's do a repeatability study And all of repeatability study is, and it sounds simple but its very important.
• 14:48
  Speaker 0: All the repeatability studies take whatever you're measuring to keep everything constant.
• 14:57
  Speaker 0: don't introduce other people or other equipment.
• 15:10
  Speaker 0: how much variation you have in that alone.
• 15:13
  Speaker 0: Because if, and you'll be surprised at the amount of variations there are... And this is where it all starts!
• 15:19
  Speaker 0: If you come back with a lot of variations like if you're consuming fifteen or twenty percent of your tolerance just by repeating them then you need to fix those first.
• 15:31
  Speaker 0: That's what I think is something else.
• 15:39
  Speaker 0: big opportunity in our industry is the repeatability.
• 15:44
  Speaker 0: And that starts with and really, The Measurement System Study and the GAGE R&R... Really what that is as a design of experiments!
• 15:53
  Speaker 0: With Design of Experiments-the first place you ought to start—and most people don't–is the Sanity Check.
• 16:00
  Speaker 0: The repeatability because if you can't repeat your measurement You're not going get good results.
• 16:08
  Speaker 0: So once you as a process statistician, one of the things I do is Howdy if?
• 16:13
  Speaker 0: If you don't have good repeatability.
• 16:15
  Speaker 0: There's two options.
• 16:16
  Speaker 0: One is to work on the actual method to try To keep it more controlled and reduce that or The second is to take multiple measurements an average them so That your weeding out in noise.
• 16:31
  Speaker 0: So that's that's the kind of work i'd does a processed statistician Is.
• 16:36
  Speaker 0: I worked with the statistics and science part, whether it be measurement system or working with engineer on a line to say something looks like changed in process here.
• 16:48
  Speaker 0: Let's go try figure that out!
• 16:53
  Speaker 1: Thank you for inside Jeff do think so?
• 16:57
  Speaker 1: You worked at different organizations.
• 17:02
  Speaker 1: Do you think that everybody has their measurement system analysis in place, or do you think it's missing?
• 17:12
  Speaker 1: Or it has to be repeated.
• 17:15
  Speaker 0: Well there is like with BI.
• 17:18
  Speaker 0: right now they have working the global animal health supply chain so there are multiple sites.
• 17:27
  Speaker 0: Everybody does things a little differently.
• 17:33
  Speaker 0: The lessons learned aren't shared between organizations or between sites.
• 17:37
  Speaker 0: That's one thing that could help because this site over here may have realized, This part of the measurement system I need to control it better?
• 17:49
  Speaker 0: Well It'd be nice if this site knew that.
• 17:52
  Speaker 0: So there are definitely a lot of organizations having measurements systems set in place to look at and assess the variation.
• 18:04
  Speaker 0: But I think, And i'm not saying this about BI but a lot of people do that as a check mark instead Of taking... This has been my experience.
• 18:15
  Speaker 0: Instead of saying I am consuming eighty percent of my tolerance just with My measurement system.
• 18:22
  Speaker 0: Well if thats the case don't Just Check it off.
• 18:24
  Speaker 0: Do something About It.
• 18:27
  Speaker 0: Does That Make Sense?
• 18:28
  Speaker 1: It makes sense, yes.
• 18:33
  Speaker 1: I also come from the manufacturing plant and there it was back in that time when i working for Corning one of the standard procedures or tasks for me to always do it.
• 18:52
  Speaker 1: For new equipment, when in your system there is something added I had to run analysis and then when i joined the research development team My first question I remember it was, do you have measurement system analysis for this equipment?
• 19:18
  Speaker 1: So i trust these equipments delivers the right values that can use and analyze.
• 19:29
  Speaker 1: or doesn't lead you to the wrong conclusions, all decisions and the decisions are super expensive.
• 19:37
  Speaker 1: And that's why I totally share these thoughts with you.
• 19:47
  Speaker 1: Yeah thank you for sharing your experience.
• 19:51
  Speaker 1: was that?
• 19:52
  Speaker 0: So companies have a lot of pressure on them making a product, characterizing it passing and giving to the customer.
• 20:06
  Speaker 0: They as data evangelist we have responsibility not just look at that data give them what they want.
• 20:14
  Speaker 0: We had their responsibilities given our advice And part of that is this measurement system has too much variation.
• 20:23
  Speaker 0: Now when you tell management They're not going to necessarily like that because it's gonna cost them more time and money.
• 20:34
  Speaker 0: See, we have to have the courage To be able to bring these up.
• 20:38
  Speaker 0: but also That's why as an extrovert
• 20:42
  Speaker 2: I'm
• 20:43
  Speaker 0: so much of me being a value-to-an organization is how good Is my rapport with the people?
• 20:51
  Speaker 0: Not just with management But How Good As My Report With The Scientist Or The Lab Technician Because That is where you get your information, by having good open honest rapport and mutual respect for the different positions.
• 21:09
  Speaker 0: I do want to say this that right here it's called AI AG Automotive Industry.
• 21:19
  Speaker 1: Yes, I used it so much when they were involved in measurement system analysis.
• 21:26
  Speaker 0: Yeah So this is the automotive group of United States.
• 21:30
  Speaker 0: This isn't older book and for SPC but They have one process themas or measurements system analysis.
• 21:39
  Speaker 0: These are the best source of information I've ever had.
• 21:50
  Speaker 0: So
• 21:52
  Speaker 1: I hope the listeners
• 21:59
  Speaker 0: will
• 22:16
  Speaker 1: remember this is a really great resource and it's written in very small, nice simple way with examples.
• 22:29
  Speaker 0: And a logical one that lays out into process of step-one or two?
• 22:40
  Speaker 1: Yes.
• 22:43
  Speaker 1: Paolo do you have any questions?
• 22:48
  Speaker 2: You do always with well-designed experiments or is it something you can also perform?
• 22:59
  Speaker 2: With the data, in your hands like in an observational fashion I would say.
• 23:07
  Speaker 2: Like because for example that i think uh...I feel a little bit same sometimes as solicitation and psychometrician.
• 23:21
  Speaker 2: So I have to make sure that when a measure fatigue in the patient, In the context of clinical trial, I minimize the error And they can detect for example The minimal amount of change That Can be associated with a real improvement Of the patients things like that and clinical trials data.
• 23:46
  Speaker 2: It's not ideal because it is collected once and you have to figure out in a way.
• 23:55
  Speaker 2: You don't have the opportunity to repeat measurement now, and then in twenty four hours things like that.
• 24:05
  Speaker 2: maybe we have measurements separated by weeks measurement variability.
• 24:15
  Speaker 2: I mean, do we have also messy data or observational data you handle?
• 24:24
  Speaker 0: That is exactly why i stay out of clinical trials because when you're dealing with humans and that's just my prerogative I really focus in on variation and controlling that.
• 24:44
  Speaker 0: With clinical trials, you just like you said can only take one measurement.
• 24:48
  Speaker 0: how do you know what the repeatability is?
• 24:52
  Speaker 0: That it's hard to know.
• 24:53
  Speaker 0: so um i really don't have any guidance for you on that other than repeated measurements.
• 25:01
  Speaker 0: uh if you can't do a repeat ability study maybe repeated measurements.
• 25:06
  Speaker 0: Uh an average those results that would help when I call weed out.
• 25:11
  Speaker 0: But that wasn't a good answer, but that is exactly why I'm not in clinical trials.
• 25:22
  Speaker 2: So most of your work is done in the context of well-designed experiments to control the process viability and measure the repeatability?
• 25:37
  Speaker 0: In the pharmaceutical world i noticed the title as CMC.
• 25:42
  Speaker 0: I'm more of a CMC statistician but chemistry and manufacturing controls.
• 25:48
  Speaker 0: But before i started three years ago in the biopharmaceutical area, I worked in nuclear ,I worked in aluminum .
• 25:56
  Speaker 0: I did a little stand-in marketing And all this was about it wasn't about bio machining to very tight tolerances, metals and composites and stuff like that.
• 26:12
  Speaker 0: And so it's a whole different industry but it shares the same measurement system.
• 26:19
  Speaker 0: problem in that is if you can't rely on the measurements in the clinical trials because you've got to be able make sure that whatever decision, whatever marker your trying to determine.
• 26:47
  Speaker 0: If you're changing something and want see if it makes a difference then can't have much of a different measurement error You know?
• 26:55
  Speaker 0: You need some things like this And typical standard.
• 26:58
  Speaker 0: what I seen is Your measurements are all of them.
• 27:03
  Speaker 0: your whole gauge R&R or measurement system needs to be, you know less than twenty or twenty-five percent of your tolerance.
• 27:10
  Speaker 0: Okay?
• 27:11
  Speaker 0: If it's consuming sixty percent of the tolerance that is a problem and need improvement in some cases may not improve but realize there really does need improvement.
• 27:23
  Speaker 0: I hope this helps.
• 27:33
  Speaker 0: so i'm sure I'm a very unique statistician on this podcast too because, uh...I am so much more about process and working to solve problems versus an end result.
• 27:55
  Speaker 0: A clinical trial or something like that.
• 27:58
  Speaker 0: but hopefully this will inspire others.
• 28:02
  Speaker 0: We can't, every person can do everything.
• 28:06
  Speaker 0: You need to find your niche and mine.
• 28:10
  Speaker 0: I found mine a long time ago thankfully...I'd hate to start over again to be honest but i found mine along time ago.
• 28:18
  Speaker 0: Find what you really enjoy doing And pursue it.
• 28:22
  Speaker 0: learn everything Ken about When I was that kid?
• 28:26
  Speaker 0: I remember I used to love the play baseball And there was a game called Stratomatic that it was before everybody had computer.
• 28:38
  Speaker 0: That's, you know... It sounds like I'm an ancient here but so it has little baseball and have cards on it.
• 28:44
  Speaker 0: they've their statistics for the year..and i really love statistics!I love
• 28:50
  Speaker 0: numbers!!And
• 28:51
  Speaker 0: thats what motivated me to be a statistician.
• 28:55
  Speaker 0: I'm not that big of a reader, but i love math and statistics.
• 28:59
  Speaker 0: And that's what I excel at.
• 29:01
  Speaker 1: Yeah So you spoke about uh What makes you?
• 29:04
  Speaker 1: Uh, what makes you special to me?
• 29:08
  Speaker 1: What makes your special is the way how You always greet me like hello my friend and I believe this kind of culture where we can be ourselves and feel free to address concerns, offer solutions.
• 29:32
  Speaker 1: It matters a lot because I know that different companies or cultures believe what really makes us ourselfs our job and what we are doing, and inspire others is being ourselves.
• 29:51
  Speaker 1: What do you think about it?
• 29:54
  Speaker 0: No I agree a hundred percent!
• 29:55
  Speaker 0: And that gets really based down to basic humanity.
• 29:59
  Speaker 0: You know if...you've got to treat others as you want to be treated ...You gotta have mutual respect.
• 30:09
  Speaker 0: um If-if you've got a bad attitude toward others then Go spend time by yourself because nobody wants to be around you.
• 30:18
  Speaker 0: You know, really it's all about.
• 30:20
  Speaker 0: even though there may be differences in religion Differences and Whatever sexual orientation or differences in whatever It doesn't really matter because everybody is a human And we all We all have our own issues.
• 30:42
  Speaker 0: even though I don't agree with you, doesn't mean that i don't like you.
• 30:46
  Speaker 0: Okay?
• 30:46
  Speaker 0: We all have to be respectful of treating people as humans because this is a hard... it's hard to live and be happy.
• 30:54
  Speaker 0: so we should be supporting one another to do that.
• 30:57
  Speaker 1: I absolutely support it!
• 30:59
  Speaker 1: And back then time as-as you uh.. As mentioned I'm also a mathematician being born as a statistician, or I didn't just become a statisticians so it's a past.
• 31:14
  Speaker 1: And i was lucky to have very good manager who really taught me statistics and had really kind personality with me.
• 31:36
  Speaker 1: You know, you have to really have this.
• 31:40
  Speaker 1: aha I understand it now feeling and that goes back through the process understanding that you mentioned.
• 31:48
  Speaker 1: And how do also with questions your curiosity about processes?
• 31:55
  Speaker 1: because as theoretical part that makes us bringing value into the organization, but it's also our interpersonal skills.
• 32:06
  Speaker 1: The environment and a lot of many things are involved.
• 32:12
  Speaker 1: I feel very privileged to work with people like you And keep being myself free to ask any questions.
• 32:26
  Speaker 1: As you mentioned, there are no dumb questions and I had an experience with one of my stakeholders when i was trying to ask questions about the process And it's just.
• 32:47
  Speaker 1: It's just what I explained right now and it made me completely.
• 32:52
  Speaker 1: I shut off Myself, and then i couldn't ask any question.
• 32:57
  Speaker 1: And I believe that they the organizations should also Bring or cultivate cultures where we feel free to ask questions.
• 33:09
  Speaker 0: Yeah, can you know?
• 33:11
  Speaker 0: What I know?
• 33:11
  Speaker 0: I don't be.
• 33:12
  Speaker 0: I does that They have that culture.
• 33:15
  Speaker 0: but in any organization, you're always going to have some people that are just... That don't have a lot of patience.
• 33:26
  Speaker 0: The one thing this sort of changes topics but it goes right along with is the one thing-the most important thing that I've found in working with somebody or hiring somebody?
• 33:39
  Speaker 0: One skill..The top skill But is are they teachable?
• 33:46
  Speaker 0: Mm-hmm.
• 33:47
  Speaker 0: Because if their teachable, They realize that they don't know everything... ...they have a dependency on others to know things and work together To get a holistic view of things.
• 34:00
  Speaker 0: So yes you're right In in that.
• 34:06
  Speaker 1: Paula do you have any input or question?
• 34:11
  Speaker 2: I totally agree and it's very important to have a curious people who realizes that we need to rely on others.
• 34:25
  Speaker 2: We are smarter, working effectively if we collaborate with them.
• 34:32
  Speaker 2: We're open for feedback and improve.
• 34:36
  Speaker 2: otherwise... it's dangerous to have people in the team that they feel like superstars of their field and there are nothing to improve on, no.
• 34:54
  Speaker 2: There is nothing left from others especially for example leadership teams where you're really pushing higher levels if you don't accept that.
• 35:13
  Speaker 2: You have to learn from others and then you have to be open, now?
• 35:17
  Speaker 2: And maybe you have two changes something in the way you operate or the when you collaborate than it's a wasted effort for the teams.
• 35:32
  Speaker 0: That is true.
• 35:35
  Speaker 0: I do want to throw this in.
• 35:36
  Speaker 0: This is a little bit of wisdom that i've gained over the years and so many times as statistician, we really have an idea like for instance measurement system where we're consuming way too much variability compared our tolerance And you can't expect a site leader necessarily share your passion because what they see is, I've got to do more QC.
• 36:06
  Speaker 0: So a good way to handle that is to work with your local Lean Six Sigma team or we call them business process excellence.
• 36:17
  Speaker 0: That's their job.
• 36:19
  Speaker 0: Their job is defined use cases that are going to add a lot of value but nobody necessarily wants.
• 36:30
  Speaker 0: Make this project up to show how much value you can save.
• 36:33
  Speaker 0: They're gonna limit it the scope so that you can go prove It out and That is what management likes to see, if for instance You say Okay I'm going have add a QC inspector which Is whole full-time employee but in doing So i Can reduce our variability of measurement enough where it's going to save me two point three million dollars in formulation costs.
• 37:04
  Speaker 0: Now, if you present that approach then they'll buy that.
• 37:09
  Speaker 0: so don't try and do all on your own.
• 37:10
  Speaker 0: reach out to local organization or six Sigma teams because thats what their value of the organization is.
• 37:21
  Speaker 1: I did want
• 37:22
  Speaker 0: say one more thing before i end this.
• 37:25
  Speaker 0: when I say when i teach and statisticians hate this, but statistics is not true.
• 37:36
  Speaker 0: Statistics Is Not True.
• 37:39
  Speaker 0: It's Your Best Educated Guest And if you take it as truth that's a mistake.
• 37:48
  Speaker 0: in bio pharma with all your ICH GMP.
• 37:53
  Speaker 0: those are all guidelines And we should use the guidelines to our advantage, okay?
• 38:01
  Speaker 0: Statistics is your best educated guess.
• 38:05
  Speaker 0: It may be right or it maybe wrong.
• 38:08
  Speaker 0: so they provide the guidelines try and guide us in making that decision but realize whether you're reading statistics are doing a statistic's its not necessarily true.
• 38:21
  Speaker 0: just your best education gets treated like
• 38:27
  Speaker 1: Very inspiring and very true.
• 38:31
  Speaker 1: We should write it down!
• 38:36
  Speaker 1: So, we are in twenty-twenty six and AI tools everywhere.
• 38:43
  Speaker 1: so this is one last thing that maybe you can touch... You mentioned a lot of challenges with data formatting really getting data prepared for the analysis.
• 38:57
  Speaker 1: So that was one of the topics we touched today.
• 39:03
  Speaker 1: And do you use AI tools for formatting data?
• 39:10
  Speaker 1: Because I've been struggling, so this is my job to clean always...to really face not-clean data and prepare data.
• 39:24
  Speaker 1: So I was given, for example PDF data for analysis and nobody has time to really prepare it for me.
• 39:32
  Speaker 1: And so um i had to take care of myself on and dig the data myself To really do my analysis and I was Saving it as excel and I saw there is a lot of merge cells.
• 39:49
  Speaker 1: This was really a nightmare.
• 39:53
  Speaker 1: I was hesitant, or really...I didn't want to use AI tools because so far um i was not very happy with the results that I get and uh because I'm working with they're really important data that really big decisions are made.
• 40:11
  Speaker 1: Um I do it myself and double check thousand times.
• 40:16
  Speaker 1: Do you think it can get better or do use any tool, or can recommend an AI tool that is suitable for that?
• 40:25
  Speaker 0: What a great question Aziza.
• 40:29
  Speaker 0: The answer to that Is You Can't Do Everything.
• 40:36
  Speaker 0: Well!
• 40:39
  Speaker 0: You cannot be A statistician, scientist, chemist and biologist a data scientist, a programmer.
• 40:49
  Speaker 0: you can't do it all.
• 40:51
  Speaker 0: You need to rely on other people.
• 40:55
  Speaker 0: we just hired somebody in our team that's a data scientists and thats her actual degree from Georgia Tech.
• 41:03
  Speaker 0: very good program.
• 41:05
  Speaker 0: she has helped so much doing exactly what your saying.
• 41:09
  Speaker 0: So let me say this I like you always try clean my data but I'm getting to the point now where i see the value she adds and I'm letting her start to do some
• 41:21
  Speaker 1: of it.
• 41:22
  Speaker 1: I
• 41:23
  Speaker 0: don't have the bandwidth or the energy to try to learn AI,to-to do this.
• 41:30
  Speaker 1: so
• 41:31
  Speaker 0: share...share the resource.
• 41:33
  Speaker 0: if you've got a big problem ,if you can fix it good .If not let's reach out to somebody to help ya.. If You Can!
• 41:41
  Speaker 0: Now that is one thing.
• 41:44
  Speaker 0: We always clean data.
• 41:46
  Speaker 0: And matter of fact, some of the statisticians like a seasoned me I know we take it as a personal responsibility to Clean data for our customer if we can because that helps win them over.
• 42:01
  Speaker 0: You know If you are trying to get your customers come on the project with you The last thing I need you to spend forty-two hours cleaning the data and sending it to me.
• 42:16
  Speaker 0: You know, that's not
• 42:18
  Speaker 1: how... That was guaranteeing you
• 42:20
  Speaker 0: would never
• 42:21
  Speaker 1: face this customer again.
• 42:22
  Speaker 1: so
• 42:23
  Speaker 0: Right!
• 42:24
  Speaker 1: So
• 42:24
  Speaker 0: you've got to be able And want to clean the data yourself To help them out.
• 42:30
  Speaker 0: I'm doing right now.
• 42:31
  Speaker 0: i do that all of time but Now..you can clean a lot Of the data.
• 42:36
  Speaker 0: excel as such A good tool.
• 42:39
  Speaker 0: It is not by far the best tool, but for basic things you can do so much.
• 42:45
  Speaker 0: So you can reach out to other people about Excel tips But always try and clean it yourself.
• 42:53
  Speaker 0: if you want to start automating things.
• 42:55
  Speaker 0: If your going to do a repetitive analysis then get somebody.
• 43:01
  Speaker 0: either learn AI or gets someone that will help you.
• 43:03
  Speaker 0: My recommendation is Get Somebody That Will Help You unless you have the bandwidth.
• 43:07
  Speaker 0: Thank you.
• 43:12
  Speaker 1: Yes, that makes sense to me.
• 43:15
  Speaker 1: and Paolo what do you have experience with that or inputs?
• 43:21
  Speaker 2: Yeah yeah definitely agree.
• 43:22
  Speaker 2: I don't have anything to add there.
• 43:28
  Speaker 0: A lot of times we try to spread ourselves too thin To learn all the new latest and greatest things And We lose our effectiveness as a statistician.
• 43:36
  Speaker 0: Aziza
• 43:38
  Speaker 1: Mm-hmm i agree.
• 43:40
  Speaker 0: so You've got a balance.
• 43:42
  Speaker 2: I
• 43:42
  Speaker 0: want to learn more, so go and learn it.
• 43:44
  Speaker 0: But is taking away from my core responsibility?
• 43:49
  Speaker 1: Yeah that's true Okay!
• 43:52
  Speaker 1: So we did... It was really insightful discussion with touched a lot of topics like from leadership From culture, data formatting Variation and measurement system analysis And many more.
• 44:08
  Speaker 1: Thank you so much.
• 44:09
  Speaker 1: Communication
• 44:10
  Speaker 0: skills
• 44:11
  Speaker 1: Communication skills matters a lot.
• 44:14
  Speaker 1: Yes, so that's our takeaway.
• 44:16
  Speaker 1: and yes or maybe we can wrap it up now at the end.
• 44:25
  Speaker 1: Can you share the secret of?
• 44:28
  Speaker 1: The power secret of a data evangelist?
• 44:30
  Speaker 1: So what are
• 44:33
  Speaker 0: those?
• 44:35
  Speaker 0: the Data evangelists to me is somebody that has a passion help others solve problems faster.
• 44:47
  Speaker 0: Because if you have this passion to do that, then you're going take put forth the effort... To help them clean it?
• 44:55
  Speaker 0: ...to get an analytic
• 44:57
  Speaker 0: ready?...
• 44:58
  Speaker 0: to help them put context into it.
• 45:02
  Speaker 0: and also You're gonna become their best friend.
• 45:06
  Speaker 0: If you take the approach where I'm a degree statistician And i don't do that then you're not going to be very effective.
• 45:19
  Speaker 1: Yes, that's a good ending and take away for everyone I guess.
• 45:28
  Speaker 1: Paolo anything do you want to add at the end?
• 45:34
  Speaker 2: It was just amazing meeting Jeff and talking with him.
• 45:41
  Speaker 2: it is always important to a knowledge that there are so many ways we can make an impact in our job.
• 45:54
  Speaker 2: So I totally agree, that uh... We don't have to stick with the latest trend but we have to fill out you know our passions and live with the best of our talent on the field where it is most useful.
• 46:16
  Speaker 0: Well, thank you.
• 46:17
  Speaker 0: Thank you so much for having me on here.
• 46:19
  Speaker 1: Yes, thank your for a group of being a great leader inspiring leader Jeff and yes And uh.
• 46:26
  Speaker 1: So it's really nice to have you hear in this episode and hopefully listeners will also learn a lot from This episode and let's keep our passion
• 46:36
  Speaker 0: okay?
• 46:38
  Speaker 0: Thank You guys.