8분 만에 깨우치는 빅데이터 혁명, 서울대 조성준 교수 강의

Ah we [music] Hello, I am Professor Seongjun Cho of the Department of Industrial Engineering at Seoul National University.

 Big data is essentially a collection of countries and data.

 If you think about Facebook or Instagram, there are a lot of people posting articles and images.

 Videos, and when  a lot of people upload such data, a huge amount of data is generated.

 Another is that the speed at which the data is generated is very fast.

 Because people all over the world are uploading at the same time, I upload and share it with my friend next to me,   so a lot of data is generated very quickly.

 Lastly, the types of data are not limited to just crosses, but texts, images, and videos are generated.

 The characteristics of big data.

 If we don't talk too technically and I'm not an IT person, what does big data mean? We can say that big data is a material for creating values ​​that we cherish.

 That value can be money, or the happiness of the people, or the peace of mind of the people, or safety.

 We can use big data as a material for creating such values.

   There are countless companies that collect data from various sources overseas, process it, and then sell it.

 In Korea, representative examples include Moi.

 Despite the fact that the e-commerce company suffered a lot of losses, it recently became a hot topic by receiving trillions of dollars in investment.

 If you think about why the company invested, it is because it has secured important resources by accumulating data on where our consumers live, what they buy, and how often they buy.

   Also, among overseas cases, there is a company called Linkin, which has the resumes of tens of billions of people.

 So, they can  recruit and apply for jobs through that platform.

 The problem was that the company barely made any money.

  Nevertheless, Microsoft invested a huge amount of 26 trillion won and acquired the company.

 Why buy a company that can't make money for 26 trillion won? In the end, the company secured more than 100 million people's data on the people who took the test.

  This is how we can see it.

 So if we just collect good data, you can think of it as a mine in itself.

 All the movies in the world are different.

 Even the movies that we remake actually follow the director, the actors, the production company, and the times, so the story is different.

 But I think the movie is different.

 So how many viewers will come to our new movie?  We don't care, but for the companies that make, produce, and distribute movies, this is a life-or-death situation.

 So, depending on how quickly and accurately they can predict consumer reactions, various sales and marketing strategies will change.

 So, how do we predict? The first step is to secure data on how similar movies to this one have performed in the past.

   We can secure data on how many viewers our movies have mobilized in the past.

 So, with that data, for example, what genre is this movie,  who is the production company, who is the director, who are the actors, what were the consumer reactions before the release, etc.

, we can  combine the tours well and create a model that predicts the AI ​​0.

 If we create a model that predicts movie viewership, what is the advantage of that? In the past, people would gather together and experts would gather and say, "This movie will probably get 500,000 viewers," and "This movie will get 1 million viewers," and so on, based on their own experiences, feelings, or intuitions.

 If a prediction is made by this, it is more objective because it is predicted with data, so objectivity is guaranteed, and it is even more accurate.

 Another thing is that when it is solved, we can figure out why it was wrong and what was wrong.

 If a person makes a wrong prediction and asks, "You said that a movie that got 500,000 viewers later got 700,000 viewers?", that person cannot answer.

 All they can say is, "I thought there would be a lot of viewers.

"  So, by using a model that has been objectified like this, accuracy and objectivity are guaranteed, and the performance of this model can be improved in the future.

 In   fact, many financial institutions are doing credit evaluations of loan applicants using big data.

 In foreign countries, not only what the person's job is, how much income they have, and how much assets they have, but also what the person talks about on Facebook, what photos they post, etc.

, it has developed to the point where it is still evaluating the person's credit rating up to 5.

 In a certain casino in Las Vegas, they evaluate the level of gambling losses that each regular customer can afford, and if they exceed that amount, they won't come back.

 So, they can't return.

 So, we  estimate the limit of how much loss we can afford with data.

 So, when we get close to that one, we deliberately go there so that the customer can no longer forget the original and rather disturb the casino so that they come back.

  [Music] 4 If you know big data, you can get a good job.

 Many companies in manufacturing, finance, distribution, and communication are desperately looking for data sites that specialize in big data.

 However, the problem is that our country's educational institutions are not providing enough of these data sites that society demands.

 The reason is that this field called data rice Essa requires studying various fields such as industrial engineering, computer panic statistics, and military statistics, so it is quite difficult for someone who majors in one field to handle it.

 I have been  running a liberal arts course called big data at Seoul National University for three years from 2017 to 10 minutes ago.

 In that course, not only engineering students but also college students, business economics students, and music students come and expel various students.

 Recently, an analysis solution that can analyze by clicking and doing it without coding has appeared.

 Since you are open source, you can ask So, if you utilize those things, I  can tell you that even liberal arts students have no problem learning data analysis.

 Recently, leading companies in the financial sector are providing big data education to hundreds of employees.

 As I mentioned earlier, liberal arts students can now do data analysis without coding, so if you study, you can utilize data as much as you want.

 So-called "  Hwa-woo-yu" can be reduced, and you can become a data scientist, which is what they say in the US these days.

  Oh, then how much do you need to do? I think that's the same as asking, "How good should I be at English?" Well, the more you  grow, the more endless it is.

 I think that the more people know, the more opportunities they have.

 When we study English, we think, "Since we're not native speakers, we ca  n't grow like that, so we shouldn't play heroes.

"   Isn't it right to say, "I can't be as good as a data scientist who majored in computer engineering, so I'm not going to learn anything?" The more you grow to the extent that you can, the more opportunities you'll have.

  [Applause] [Music]