Welcome to this presentation about Data Science which is an important - - PDF document

1 Welcome to this presentation about Data Science which is an important topic in the master of Business Information Technology.

2

3 Scientific and economic progress is increasingly powered by our capabilities to explore big data sets. Data scientists dig for value in data by analyzing for instance texts, application usage logs, and sensory data. They are the driving force behind the successful innovation of Internet companies like Google, Twitter, and Yahoo. There is an increasing need for data scientists and big data engineers seen in job

• advertisements. The need for data scientists and big data analysts is apparent in almost

every aspect of our society, including computer science, biology, medicine, physics, and the humanities.

4 For some time, I have been collecting job opportunities mentioning “data” or even “data scientist”. Almost all of our students have a job within 3 months after

• graduating. It is not uncommon that students have already secured a job even before

they graduate. In short, no need to worry about finding a job: the job market is very good for data scientists.

5 But more importantly, data science is very interesting and cool. It is a kind of magic how seemingly intelligent behavior may arise from a sea of data. You can be a magician! Let me explain what I mean by that by giving some examples.

My favourite example is this one. Imagine you took this picture during one of your vacations: a nice lake surrounded with hills and pretty villages … and an ugly building in front. Recognize that? 6

You can of course try to get rid of the ugly building using tools like Photoshop. We cut

• ut the building, but now we have a hole in our picture. Now what?

7

The researchers of the mentioned paper produced an algorithm that can be used to fill in the empty part with something relevant. For this picture it produces the result on the right. Nice boats. A bit of suitable rippling on the water. A bit of tree in the bottom right corner … How on earth can an algorithm know that this is a relevant filling for this hole? Isn’t it magic? 8

Well, the algorithm itself is not that complex. the main trick is that it uses a collection

• f pictures. It doesn’t work with 10.000 pictures, It doesn’t work with 100.000
• pictures. But all of a sudden when going beyond a million pictures, the algorithms

starts to be able to find relevant patches and pasts them in producing results like this

• ne. Of course there were not boats on the lake when you took the picture. But you

can’t deny that this is relevant content and it really solved your problem with the ugly

• building. Isn’t it magic?

The paper is from 2007. Can you image what we can do nowadays, more than 12 years later? 9

My second example is from 2011 when IBM had their system “Watson” compete with against former winners of the quiz show “Jeopardy”. The game show host would ask questions of any kind and the contestants need to answer them. It is not very magical that Watson won: a computer can press a buzzer much more quickly than a human

• can. But the magical part is that the system seems to understand the question in

natural language, would know general world knowledge, and could answer in natural language. Since when can computers listen, understand, and speak in natural language? 10

And can you imagine the business potential of computers being able to read? It is as if you can have an army of lepricons at your disposal to, for example - you being a clinical - sift through and read all letters and reports in the dossiers of millions of patients looking for similar cases or gathering statistics for you. Wouldn’t you want to understand this technology and being able to design services that harnesses this power for the good of our economy and society? Wouldn’t you want to be a magician? 11

I like to explain what data science is, by explaining what process the data has to go through, because these are the activities you need to master to be able to exploit the power of data science and AI. First, the data comes from a variety of different sources: organization’s information systems, sensors, web pages / fora / applications on the internet and in particular social media. Relevant data needs to be found, harvested (extracted from the sources), combined, transformed, cleaned, etc. for it to be usable for analysis. Analysis ranges from reporting and visualization to obtaining intelligent predictive models using machine learning and data mining. Finally, to be able to successfully use these analytical results for decision making and smart services, one needs to be able to interpret and evaluate them. Having the necessary skills on each of these steps will allow you to design and realize smart services that harnesses the power of data for the good of many domains in our economy and society: health care, logistics, maintenance, sport, etc. 12

13 Industrial innovations will be increasingly done by analyzing big data.

Perhaps you have heard about hypes like Smart Home, Smart Industry, etc. What do you think is responsible for “Smart” in all of these … ? Of course Data Science! Let me give another more serious example: health care. 14

This is a quote from a data science professor. 15

There is so much potential for using data science and artificial intelligence to improve

• ur health system! Please find some examples on the slide.

We need the data scientists to re-think the business processes enabled by newly designed smart systems and services. In fact, I believe that we will experience a day in the near future when we do not go to

• ur doctor when we are not feeling well, but that we receive a message from our

doctor saying something like “Tomorrow you will not feel well. It is nothing serious, you need not come by to see me, just wait it out for a few days; I’ll monitor your illness remotely and whenever there is something to worry about, I’ll contact you again”. Much better diagnoses and treatment decisions, no more unnecessary visits to the doctor, no more unnecessary use of resources, etc. You can be the magician that can help to turn the whole health system upside down to improve it. 16

17 Also science benefits enormously from data science.

18 Jim Gray is an ACM Turing Award winner (i.e., the “Nobel Prize” for Computer Science) He predicted in 2006 that science would enter a new phase, which he called the 4th paradigm in which new discoveries would be primarily driven by data analysis. To explain, the others are: 1st: empirical (conducting experiments), 2nd: theory (such as Kepler's, Newton's law, Maxwell's equations), 3rd: simulation (to be used when analytical solutions based on theory become impractical); and the 4th is: analysis of Big Data. This is where IT meets scientists. Data Science is needed when “My Excel Spreadsheet gets out of hand”: What happens when you have 10,000 Excel spreadsheets, each with 50 workbooks in them? Being able to integrate and analyse data on a large scale has the potential of answering scientific questions that cannot be answered using the other three paradigms. Tansley, Stewart, and Kristin Michele Tolle, eds. "The fourth paradigm: data-intensive scientific discovery." (2009).

To give an example of how data science revolutionizes science, think of the social

• sciences. Studying for example human behavior in the work place and in professional

teams is traditionally done by asking people to fill in questionnaires before, during and after a specific task. Nowadays we put sensors (such as sociometric badges) on people and observe them with microphones and cameras while they are doing their work or work on specific tasks. And then analyse this data. This gives a much more continuous view of what is happening. And the observation is less obtrusive, subjects can function without even being constantly aware that they are being observed. But social scientists are not data science specialists. They need systems to support them … and BITers are typically involved in designing such systems. 19

Beware of the power of data science and AI. Obviously, you will be taught to think about ethical concerns and how to properly address them in your designs. 20

21 And obviously a currently very visible application of data science are websites like these gathering data on corona world-wide and analyse it in real-time to produce a dashboard like this. A BIT master student would be able to design and realize such a system. Source https://www.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b 48e9ecf6

22 22 The Master BIT has two specialisations: (1) IT Management & Enterprise Architecture and (2) Data Science & Business (DSB) In both you will learn about the basics of data science, but if you choose the DSB specialization, you will study in depth about these 4 topics.

23 23 Within the DSB specialization, there is room for choosing your own courses, so that you can further specialize and accommodate your course program to your own

• interests. You can both further specialize on the business side as well as the technical

side.

24 24 The University of Twente does not have a separate Master in Data Science, but you can study Data Science in Twente by enrolling in one of these four master programs and then choosing the associated data science specialization. A further possibility is to study Data Science in the EIT program which means that you study one year in one country and one year in another. You can choose UT for both the first (entry) year as well as the second (exit) year.

To help you choose which data science specialization is the appropriate one for you, look at this picture. On the left are different kinds of data: ‘normal’ tabular data (spreadsheets, databases), textual data, audio/video data, and sensor data. On the bottom you see the phases of the data science process I mentioned earlier. You see that Computer Science focuses most on handling tabular and textual data, Electrical Engineering focuses most on audio/video and sensor data. The DSB specialization of BIT is focusing less on collection and data preparation and more on deploying and using the technology in smart services. Applied Mathematics focuses on the theory and algorithms behind machine learning and analytics. These focuses are not absolute: it is a matter of emphasis. All specialization share some courses and it is allowed to choose elective courses from other studies. 25

This about what you will learn … what kind of careers would be possible for you with a BIT/DSB study? 26

You could be an entrepreneur … and then become a multimillionaire after a couple of years just as these former students of ours who started the company Distimo and sold their company to App Annie several years later. Three of them are BIT students; one is a computer science student (Tom). 27

Or you can choose for a career in science. Here you see Meike Nauta, who studied a bachelor BIT first, then did the data science specialization in Computer Science and is now doing a PhD with me. She won several awards with her bachelor and master final projects. 28

29 You can even become a politician ;-)

30 During one of the courses (managing big data), the students participated in a big data challenge …

31 … and these three students won it! One started a company and is now a data scientist at a company of another former student of ours. One is a data engineer at an energy infrastructure company. And one went for a PhD first and then decided to continued as a freelancer, when he found his financial circumstances not needing work-for-money ;-)

Another area where many data science students end up, especially BIT DSB students, is consultancy at the various large and smaller companies. Here are a few examples of former students of ours. 32

33 If hope you have a better idea now of what data science is, what interesting and important work data scientists can do for our society, what you will learn in our data science specializations, and what kind of career you could pursue. If you have specific questions, then please contact me or my colleague Christin Seifert.