We went on a deep dive on what it takes to be a data-driven workplace and how you can make smarter, data-informed decisions in your everyday work! We also got the chance to answer some of our audience’s burning questions about skills you need to become a data analyst and cleared some questions about our Data Analytics courses.
We had a blast, enjoy!
Academy Xi (Olivia Bowden, event host): Hi JP! Welcome to our fireside chat series. We are so excited to have you with us today. I think I’m gonna hand it over to you to introduce yourself if you don’t mind, let us know a little bit about you, your background, and your journey towards the world of data.
JP Hwang: Fantastic, yeah, thanks Liv. So, my background at university is in engineering and in IP or intellectual property but then I started working on a tech start-up which was all about extracting competitor information and market trends using scientific papers and patent data which is quite information rich and dense. In that process, I was immersed in data science, machine learning, and programming and all that good stuff.
More recently though, I’ve kind of identified a need in a lot of organisations, where, for many of them, if not most, the volume of data really outweighs the analysis tools and just general capability to do something meaningful with it.
So, I’ve kind of pivoted. What we’ve been doing now is to help organisations make better use of data to help not only analyse it but to better consume the outputs and make better use of it going forward.
Some of the stuff we do and what we have done is things like help with the internal knowledge management systems by building databases and apps to access them and manage them over time. And things like business intelligence apps for analysing customer feedback or customer feedback on various products and groups. Or even doing cool things like analysing video feeds to extract data information in real time, so you can provide real time feedback.
I have been working with clients small and large from start-ups to fairly big companies and, more recently, obviously we’ve been working with you guys at Academy Xi to create the Data Analytics Course. It has been a really rewarding experience empowering people in that space as well.
Academy Xi: I remember when we first got to know one another, I had a look at your LinkedIn profile, you haven’t always been in data, this is something you’ve been moving towards in the past couple of years. Do you want to take us back a little bit and sort of tell us a little bit more about your journey?
JP: Data science is an interesting field, because it’s a new field–it kind of combines statistics and computer science. It’s a new discipline so I’ve come from an engineering background, so I’ve got a technical background. I’m used to coding all my life basically. And then I worked as a patent attorney for a few years as well and that’s where I kind of the two worlds came together in terms of programming and the IP space and I was working on that start-up right, so it was kind of all about leveraging the gigabytes or terabytes of patent data to do something with it and make it easier to extract intelligence and information out of that.
And that’s the start of the formal, I suppose, move into the world of as a developer, and as a data scientist.
That’s been unexpected, but it’s also been really fun and rewarding because it’s a new field; there’s a lot of exciting developments going on and I’m sure you’ve all heard the discussions going on about AI. You’ve probably seen bots, you know chatbots, or whether it’s language generating bots. Even these days you’ve probably seen pictures from things like DALL-E that generates images based on the prompt you’ve given them.
And there’s a couple of things really cool about that, which is learning how to use these tools which we’ll talk about a bit later. But the creative prompts that people give those AI bots are really interesting because there’s an art to using them in terms of giving the prompts the right way.
Academy Xi: What I think is great about your story as well, and I think a lot of people on the chat will resonate with is that you haven’t always been a data scientist. That it is possible to transition and take those transferable skills and carve a career out for yourself in data.
But we’re also going to have a lot of people on this call, presumably like myself, who aren’t too overly familiar with data at all.
I guess the first question that comes to my mind is, why do all of us need to know or care about data?
JP: I think that’s a good question and over the last couple of years, the word data has taken on this kind of a fairly mythical and intimidating quality. So let’s just kind of talk about what it is right?
I guess the more boring definition will be that data is just information that’s been collected and quantified. That is any kind of information that’s been written down essentially or recorded in some way, even if it is not particularly meaningful or necessarily even important nor you might not see what the hype or the fuss is about.
But I think, where we can all agree, though, is that being able to understand the world around us better and making higher quality decisions is really, really important and desirable.
Data allows us to do exactly that and the rise of things like Machine Learning and AI in terms of having bots, for example, that can look at our radiology pictures and come up with a really accurate diagnosis that outperforms physicians is all enabled by data.
So it’s really those enablers and the way to leverage data to be better informed, to make better tools and to make better decisions. I think those are the reasons to really care about data.
Academy Xi: But in the enterprise side of our business, we have been hearing about how we can make our people or allow people or empower people to make more data driven decisions. Have you got any examples you can think of, in a corporate context where people could be using data just to do their everyday tasks a little bit better?
JP: In professional capacities, if you’re doing things like capacity planning, you’d be looking at things like your seasonal and annual data from your historical perspectives and you’d have data about what your product release cycles might look like; and how that affects your capacities.
If you’re an R&D manager in a tech firm or a research firm of course, you could look at scientific publications and patents to get trends and see what your competitors are doing.
And one of the biggest changes in the last 10 years has been doing things like AB Testing in a lot of domains right? That’s websites, that’s marketing–and that’s extremely data driven in that we can do that. To be able to systematically analyse it to make better decisions on what people want: what kind of interfaces work better and what kind of workflows work better for our customers.
Academy Xi: Yeah fantastic. It would be amiss for us to have a data expert here and not actually talk just briefly about what the data analysis process looks like for somebody who’s a specialist–say a data scientist, a data analyst? Can you explain what that process is all about, and then we can drill down into other areas?
JP: Yeah sure. So, Data Science and Data Analysis is such a big field, I won’t pretend to kind of encapsulate all of that in one go. But I can talk a little bit about what data analysts do which is, I guess, a little bit more defined in terms of using data to answer organisational questions. And to come up with analysis that can be leveraged directly in an organisational perspective.
Typically, the process begins from defining the problem for the analyst and that might be internal or external, meaning translating the business goals and objectives to specific data driven analysis goals, and these are typically SMART goals right? In the same way that we set SMART goals for any other things that we do.
Move forward then of course to the process of analysis which I won’t go through in detail because it’s quite dry, to be honest. It’s all the steps of gathering data, cleaning data, exploring and then finally performing the analysis to come up with insights from our data.
And the last step is really the key, one that non-technical people or non-data professionals would really engage with a lot, is really making use of the data and effecting change. That might be communicating those outputs to the stakeholders to help them make better decisions or it might be you consuming those insights yourself to act, or just be better informed about the world around you.
That’s a really long way of saying that the Data Analytics process is all about answering questions about our world with data so that by the end of it we give ourselves a better chance of making better decisions than if we hadn’t done that.
Academy Xi: Yeah, fantastic. In terms of actionable insights and giving people tips on how to do that and I understand from what we’ve talked about, there are three key pointers that you’re going to be taking us through today?
JP: We’re going to start by talking about developing your data intuition. For a lot of non-data professionals, the key part of your data literacy skill that you should be focusing on is really the reading and comprehension part of data literacy. What that means is being able to understand, internalise, and make use of data driven outputs right. So, as we’ve been talking about the volume of data collected and data driven outputs, many of you might already be dealing with these outputs. So you probably heard about reporting terms like ‘confidence intervals’, ‘margins of error’ and then all the talk about what a model can do.
The good news is most of you will never need to calculate these statistics or develop any models yourself. What you will be doing though, going forward is to be asked to consume these outputs and to integrate them to your work. That’s why developing your data intuition, which is a gut feel for what those data points tell you, and how you can use them is important.
Academy Xi: This is super interesting JP and I’m just going to pause here to say we had one of our attendees pose this question: “Hi, I’m a marketer wondering about instinct or gut-feel versus being data driven, do both have a role to play in better decision making? When do you need a balance of both?”
JP: What people think of gut feel or instinct is interesting because that’s basically your brain having internalised patterns that you’ve seen. So, it’s not really codified in terms of hard numbers but there’s quite often something to that. So how those insights and gut feels, or those intuitions get used might have been a process that you would form some hypothesis about your world, what’s happening with your customers, and what have yous using those gut feels.
And then, what you might do is to develop a Data Analytics process based on it to test the hypothesis right. To see what is happening. How likely is it that it’s happening if you make improvements, based on your findings.
How you leverage these insights and quite often, of course, it turns out to be not true because we all have internal biases.
Academy Xi: To that person, say we thought about a customer service team manager, so how, in your opinion, would this kind of person develop “data intuitions” in their role?
JP: Yeah, that’s an interesting point, and of course it depends on every role. I won’t pretend to be an expert on facility managers, forgive me if the example’s not perfect.
But, as I was thinking about it, I think it would really help them in designing the workflow in terms of how to use data and how to engage with people. So, let’s say a customer service team manager has some thoughts about how to improve their workflow for the customer. Knowing what types of experiments to conduct to find out more about this, how many people, they’d need to survey or get samples on whatever how much money it costs as a result. And then once you get the results back. What you do with it and do what you know to interpret it so they can make sense of it.
But then, if you just had a general sense of what’s possible, once they understood what AB testing is and how it worked and how those two things you conducted and who to talk to will really help them get started.
It’s the same thing with integrating things like chatbots, which is, I think, is used a fair bit in the world of customer service. We kind of see that a lot on websites and quite often you’ll see that pop up and it’s not a real human, you’re just talking to a chatbot.
Chatbots are cool but they can’t do everything right? So, if you understand the kind of things they can do well, which is things like, question and answer, and retrieving certain parts of information they’ll be able to then look at their data and say, “well how much work for the representative is that? How much would introducing chatbots help to our workflow? How much would it cost? They can do that cost benefit analysis much, much more easily rather than just if you didn’t understand what chatbots did or couldn’t do. It’ll be hard to make that comparison and think about whether it’s worth integrating tools like that to your work.
Academy Xi: Moving on to the second one, because this for me is always a fascinating area. I think we’ve all been privy to some dodgy confusing data visualisations in our time so dealing with data visually this is point number two.
JP Hwang: I do a lot of work in data visualisations just because it’s I think an essential tool for those who work with data, kind of day in, day out. But also, for consumers of data, the thing is that we simply consume a lot more information and faster when it’s visually encoded. If you had a table and tried to make sense of any of it and what’s the pattern and there’s thousands of rows and that, but if you see a graph, you can see patterns on that clearly, so that’s one thing one can benefit about data visualisation.
Another good thing about visualisations is that unusual patterns reveal themselves quickly amongst thousands of data points, and this is a data and looking at.
Academy Xi: I guess at the end of the day, this is about clear communication and setting yourself up, I guess, with some simple tools to help you do that, which links us to this question from our audience: “What software tools can one learn to be an expert in data visualisation?”
JP: There’s a few depending on your workflow and how you use data visualisation. Some of the more popular ones like Tableau, Power BI where you’re using desktop software. And then there’s online tools like DataWrapper or what’s called Flourish.
For someone like me, I obviously do a lot of my work in the programming space, I’ll use programming libraries like Plotly or Bokeh or even something called D3. So it’s really horses for courses in terms of what you’re doing and what you’re trying to present.
In a lot of cases, if you’re just creating charts, simple charts for presentation purposes, something like Excel is really good enough because it’s not really about creating the most fancy charts and the most complex charts. But it’s about really communicating your message through whatever means necessary.
Academy Xi: Lets loop back to that customer service team manager. How would you envisage they use visualisations to communicate data clearly?
JP: There’s two sides of that right. So as consumers of data visualisations it’s important that they be able to communicate with the stakeholders or service providers who might be creating these visualisations, so you understand what it is that they’re telling them.
Quite often, though, depending on where you’re at in an organisation or what your role is you, quite often will be in the role where you must communicate your findings from your studies to stakeholders higher up.
That means selecting some charts or even creating some charts yourself to communicate your message and key takeaways to your stakeholders. In that case you’d want to be able to make at least some charts and it’s not probably as intimidating as it feels for a lot of people to learn how to create these charts.
And, as I mentioned, those online tools like Data Wrapper or Flourish that I really like, are very easy to use. If you know how to use Canva, you can probably use a lot of that too.
It is about empowering yourself and as well as others in the organisations with tools, right? There was a stat that I read, which was mind blowing and it said that “when people are faced with data-related tasks about half the people would either avoid the tasks or not use the data provided at all.”
So that’s quite interesting. And on top of that, about three quarters of the same respondents said that those tasks made them unhappy or overwhelmed. That’s a really striking stat for me because that means for half of those people, they’re not using the data. So they’re not any better off than they would have been if they hadn’t been provided with the data.
But now they’ve been left kind of disenfranchised or unhappy, and disempowered–sort of feeling like they’re not capable of doing this. Data, for those people, made their lives and work worse than if they hadn’t had that in the first place.
One of the key recommendations here is to kind of really empower yourself with a tool that can add value to your work, and it doesn’t have to be anything super fancy like a programming language. And it can be something basic like just learning how to use a couple of formulas in Excel or load up data in Excel or Google Sheets, it might be something like Tableau. Or even those online tools like drawing charts in Canva or Data Wrapper, as I mentioned, and that’ll be really quite empowering.
It depends on what your work is and what your relationship with data is, but I think it can be a kind of transformative experience.
Audience question: What is the difference in job profile between a data scientist and data analyst? If there is a difference, what should a data scientist do in the data analytics field?
JP: One of the things that we wanted to do at the start of our data analytics course was to disambiguate between these two terms between data analyst and data scientist.
The truth of the matter is that they are fairly nebulous terms, but I’ll try and do what I can in terms of the best of my understanding.
Data science is generally a broader term than data analysis. We talked about the process of data analysis in terms of problem definition, all the way out between data cleaning, data managing and analysis, all the way to delivery of data communications.
Data science tends to be a little bit broad, broader (excuse me), and a lot of what data scientists do is things like develop models. You know if you’re talking about people who develop state of the art models in image recognition. Sorry, I’m getting fairly technical. Like if you see those bots or a website where you’re putting (uploading) a picture and it tells you what it is.
Or, if you’re driving a Tesla and it’s able to drive around and figure out if they’re driving on a road or if they’re about to run into another car or follow lanes and stuff—these are all driven by AI models developed by data scientists
That’s probably how I understand data science and data analysis.
And in terms of what should data scientists do to get into a data analyst field, I think some of the skill sets are slightly different. Data analysts tend to work with things like SQL a little bit more. They’re dealing with kind of more predefined roles and analysis pipelines because they’re more often in a business role answering defined business questions with these tools.
Whereas data scientists’ roles, like building models so it tends to be a little bit different from that sense.
Audience question: We often hear from L&D managers, ‘look, our executive has said, our people need to be more data literate and make more data driven decisions across the board, where do we even start? People have varying levels of competency already, how do we even begin on this journey?
JP: There’s a couple of things that can do that can help, obviously, one thing is just kind of reducing that intimidation factor in dealing with data. I think a lot of that is going to be just training and continued development for those people and supporting them throughout.
In terms of how to figure out where they’re at in terms of their comfort with data and abilities, I think it’s a good idea to have conversations with people about how comfortable you are in dealing with these tools that might be useful on top of or in your day-to-day tasks.
And then you can ask them questions like, “how comfortable are you with Excel, formulas, pivot tables and VLOOKUP formulas and so on and so forth?”
It might be that you’re dealing with databases all the time. How much do you use SQL you know? It’s probably quite intimidating to be formally assessed in something, but I think a lot of training would be a really good way forward for many organisations and it can be a collaborative process to put them in groups and work on kind of relevant business problems together.
Audience question: Do data scientists need to have knowledge of statistics and what level of it do you need to become a data analyst?
JP: I guess by nature if you’re dealing with data science and data analytics types of fields, it is undoubtedly going to be computer heavy so levels of computer literacy is really helpful.
But I guess the question is, can someone without an IT background join the data world, I think the answer is yes, none of us are born with these skills, right? These are not innate skills that are encoded to our genes or anything so for sure it is something that we can learn.
And I’ve seen people do it in the Data Analytics: Transform because a lot of the students in our cohort that I’m mentoring come from a non-technical background, I would say, actually, most of them come from non-technical backgrounds. And a lot of them pick it up really well and they do fully admit that it is not easy, but it’s something that they pick up and they do really well so that’s great to see.
In terms of statistics, there’s two parts of this question I can see, so I’ll try to answer the first part about the knowledge of statistics. Depending on what you’re doing in data science. If you’re building models and using machine learning to do things like regression models which is when you predict things like numbers, so when you predict things like housing prices, based on where it is, how big the houses are, blah blah blah. Or if you’re using models at any level you want a fairly good intuition of statistics and maths.
But if you actually work on some of these fields, what you’ll see and what’s required isn’t a university graduate level of statistics or algebra or calculus. What you do need is basically a low level understanding of some statistics and some algebra and that’s kind of all you need.
If you of course, are doing some role that is fairly statistics heavy and that might be in the fields of experiments where you are required to do things like, “does this experiment tell us with a certain degree of confidence that this drug is going to work”–or something like that those fields are, of course, more statistics heavy and critical, so it really depends.
Audience question: Is it worth doing a master’s degree in this field or would the Elevate course or Transform course offered be enough to become a data analyst or data scientist?
JP: That’s a really good question. I don’t feel particularly well placed enough to say whether masters is worth doing as that’s a fairly personal type question.
The Data Analytics: Elevate course is designed for primarily non-data professionals trying to upskill and sort of become more familiar in dealing with data and to empower them as we mentioned and to give them a flavour of what a professional data analyst might do.
That’s why we have optional modules and things like using Python, which we encourage people to look at, but it’s not part of the assessment or anything like that. And we do the same thing by giving them a flavour of things like machine learning and to show them what they can do.
The Data Analytics Transform course is meant for exactly that, for people who are trying to get into the field of data analytics. I guess that’s the intent of it. It’s an intensive course and I think it’s designed to be around 15-20 hours a week, you know for 14 weeks part time. That is what it’s what it’s intended for, and it covers the gamut from learning how to use Python. Using statistics and ending up with being able to build regression models, as the capstone project, where you can pick a subject of your choice or topic of your choice build regression model out of it and be able to really complete that loop, I suppose, from problem definition stage to driving some insights and communicating it out to external stakeholders.
Audience question: If you don’t have a data analyst in the business it’s hard to get the data you want extracted, any tips on how to implement a data analyst into an organisation?
JP: I think it goes twofold: I think it’s important for data analysts to understand what the business objectives are, what their priorities are and how they work.
And conversely, it’s important for the organisation to understand what data can tell you, and what the data analysts can do for you and what they can’t do. So, understand the limitations. That’s because it’s important for data analysts at the problem definition stage to clearly understand the objectives as far as what the business objective is. So that they can translate that into actionable goals.
If that doesn’t happen necessarily well, what happens is that the business stakeholders would provide a brief and that necessarily doesn’t get translated into a data analyst’s understanding. They go off and do the work, get some results back and it turns out that it doesn’t necessarily answer the question that they were looking to answer; or it’s not particularly actionable.
And from the data analyst perspective, quite often the complaint I hear is, “well they asked me to do this thing that’s not actually possible so now, I have to manage their expectations about what is actually possible with data, as well as to try and answer this question some way”. So if they have an understanding of the limitations of data and analysis they can obviously integrate the data analyst or team better into the organisation and make better use of that resource.
Audience question: Do you have a structure when communicating the insights you’ve garnered from data to stakeholders?
JP: Yeah, that’s a good question. I think of data communication as similar to a part of rhetoric, so when you’re trying to convince someone or change their minds on something, the data should really be designed on backing that up.
A good place to start would be to think about what your goal was when you started the data analysis and kind of anchoring everything back to that.
In terms of this structure or the methodology it doesn’t have to be something complex, so you might see some beautiful data visualisation online or on Twitter—I see and I follow a lot of these people—so I see a lot of that, but then you kind of look at it and even for me who is quite used to them, I kind of go oh that’s really pretty but I have no idea what that says, but it’s really pretty.
So sometimes just a table with like three numbers might be better than a very complex beautiful data visualisation because it tells a clear message.
So that’s my way of saying that, whatever does the job is good. It always is what you want. And, I know I said less is more in terms of visualisations but it fits for the right audience, you can actually do more complex things.
For certain clients of mine, I’ve built them some dashboard apps that look at data that connects to their data pipeline. But what that allows the audience to do is to dig further into the data themselves. Those types of outputs tend to be a little bit more complex but because they have the expertise in their domain and because they are a little bit more data savvy, they’re able to dig into the data themselves. They won’t be performing the analysis and cleaning the data and so on and so forth, and building those models; they can look into it and answer their own questions as they go.
Audience question: I am a junior marketer and hoping to be able to make use of marketing data to optimise marketing decisions and campaigns, however, I don’t come from a technical background, so I wonder which course I should take in terms of elevate versus transform?
JP: I think, for me, the Elevate course would be preferable in your use case. It is designed around non-data professionals looking to upskill and really understand more about the world of data and data analytics at answering questions using data. That’s my two cents on that.
Of course, if you’re looking to challenge yourself, I wouldn’t discourage you from taking that Transform course, but my recommendation is to do the Elevate course.
In the Transform course we do cover a little bit more of things like statistics and hypothesis testing and A/B testing and how to analyse results of that more clearly. If you’re specifically interested in those areas, you might look into the Transform course. It is significantly more challenging because it is programming heavy.
Audience question: I have a background in tech sales but am quite curious in terms of how I can use data. I’m thinking of reaching out to our marketing team in house or other departments to see where I can add value. Any tips on what tools I can leverage to work on during my own time?
JP: A lot of things that people do is try and build their online portfolio in data science and analysis and there’s a lot of data out there that’s good – really, really high quality and publicly available. What people do is to source some of that data in whatever domain you’re interested in and build a portfolio of analysis and build your publicly available profile.
Now you can do a similar thing with your organisational data probably too depending on what is available to you, given your role and how accessible that data is. It is always always useful to solve a problem that the organisation has and quite often what they’ll have is things like there’s a lot of data in this field and they’ve collected it, for whatever reason and they probably haven’t had time to do anything with it, and so they can probably give you a background on what they were thinking when they collected this data. And why they haven’t actually done anything with it and it might be something as simple as they have collected all this data but it’s quite messy.
So it needs someone to go through that and clean it and that’s a really valuable process. There’s a running joke that people think data science is extremely glamorous and you’re building these AI bots and whatever. What you’re actually doing is just cleaning data and getting rid of typos like 90% of the time and that’s more true than people actually think.
So, you know I would say talk to people who have data in your organisation and talk to them about the background, as to why. Talk about what their needs are, and you can probably help them to address some of that and it’s a good way to build your and to network as well with the parts of the business.
Audience question: How do people in an organisation best work with specific data teams? There are often teams of specialists who are excellent at what they do, but how can others in the business really tap into that resource in an effective way?
JP: I think it’s about engagement and about data literacy. I think quite often what you see in organisations is that perhaps they’re a little bit too siloed in terms of what they do.
So really engaging with them, talking about what they do, and then that also helps them understand what you do and what your needs are as well.
I’ve spoken to people who are doing things like rolling out tools like Tableau in the organisation. They said getting to engage throughout the organisation, because they were rolling out these tools, helped them to speak to different parts of the organisation that they otherwise wouldn’t have to.
Kick start your career and gain in-demand skills with our 100% online Data Analytics courses.