Catch data big and small! (Overview of Data Science courses from the Cognitive Class)

Recently, increasingly I come across the mention of "Data Science" or, in our opinion, "Data Science". I am not a specialist in the field of IT and throughout my life I am not friends with the mat. analysis and statistics, so I have long gone past this issue and probably would have continued to pass by, but at some point curiosity took over.

So, Cognitive Class , also known as Big Data University from IBM (sometimes abbreviated as BDU) is a portal with free courses on topics close to BIG Data and, accordingly, Data Science.

Want to know what he can or can not teach you, then you are welcome under the cat.

So that you have an understanding of what eyes I looked at this course a couple of theses about me:

1. Not friends with mate. analysis and statistics;
2. I know how to code a bit (write “code catches” and somehow working “programs”);
3. At the very least, I understand English. Written - tolerated, by ear - bad.
4. Previously faced with on-line learning.
5. At the time of registration in the Cognitive Class, Data Science did not know anything at all.

I hope that you have read my short biography for good reason and it will still come in handy.

We will proceed from the fact that people with different “backgrounds” will look at this course in different ways, so without pretending to be objective, I’ll start:

Part 1. Why Cognitive Class?

By ridiculous chance , I am deeply convinced that there must be a lot of other high-quality and useful material on this topic, it was just one of the links and I went over it.

It should be noted that in the Russian-language Internet space, neither under the new brand (Cognitive Class) nor under the old (Big Data University) portal is particularly “glowing”. Most likely the main reason is that it is not translated into Russian.

Nevertheless, the fact that all materials are free , most of the courses are based on open- source software , and at the end give out some certificates and badges (about them later), in combination with curiosity did their job. Well, as a plus, you can practice English.

Part 2. Batman Data Science: The Beginning

After registration, the site provides access to a variety of courses. All the courses that I came across could be started at any time, there was no time limit, interaction with teachers or students was also not required.

Each course can be taken separately, or as part of the learning program. An electronic certificate is issued for each training course, a badge is issued for fulfilling the requirements of the training program.

The site interface resembles any other distance learning system, so I think the process itself should not cause problems for experienced users.

Since I did not know anything about Data Science at the time of registration, God himself commenced to start with the Data Science Fundamentals training program, in principle, this was confirmed to me by the local Skynet. The site has a bot (Student Advisor), if he writes: "Data Science", then he just adds this training program. The bot is not suitable for any more complex and sincere conversations, because it only seems to understand the keywords from the topics of the courses.

Let's get started On the curriculum page, it is clear that it consists of several courses, ranked in the recommended order of passage (although no one forbids to go in any order).

At the same time, to obtain a first degree badge, as a rule, it is necessary to master the very first course of the curriculum; to obtain a second degree badge, it is necessary, as a rule, to complete all courses of the program. Consider it in more detail.

Part 3. Data Science Fundamentals

The program consists of the following courses in the recommended order of passage

• Data Science 101 (first badge)
• Data Science Methodology
• Data Science Hands-on with Open Source Tools
• R 101

At the end of all - the second badge

Briefly about the concept of "average rate" on the platform.

Each course that I came across has the following properties:

The main material is divided into modules (it looks like there are always 5)

There are auxiliary sections such as introduction, exam, reviews.

In each of the training modules, as a rule, there is one video (more often) and almost always a set of test questions on the material studied, there are also usually laboratory tests and articles sometimes come across.

The rating scheme is usually 50% / 50%. The first half of the final tests of the modules, the second for the final exam. Passing threshold is usually 70%. In some courses, the values ​​may vary, do not be lazy to look at the Grading Scheme section.

Watching all the videos or doing all the lab tests is not necessary, as long as you pass all the tests and pass the final exam.

The tests themselves are quite simple, for any questions, except those where you need to poke true / false, two attempts are given without penalty (in true / false - like a sapper there is no margin for error). Usually at the end of the module there are 3-4 questions, most of them with one answer from 3-5 values, sometimes there are questions with a tick, sometimes with a field for input, questions usually about the material and simple to the point of horror. Time to pass the test is not limited, you can "poke" in the answers at any time.

Unlike the tests, the test must be passed in exactly 60 minutes (but possible earlier), otherwise it is similar to tests, only more questions (10-20).

At the end of each course will give a certificate and if provided for a badge (about them closer to the end)

Let's briefly review each course of the program, well, and I will share my impressions of him:

Data Science 101 - Useful, inspiring course, something like the course "How to become a programmer in 2 hours from well-known programming schools." In the first video, the venerable scientist from Canada, and the young guys will tell you who Data Scientist is and what they eat it with.

Let me remind you once again that all the courses that I caught were only in English, but in most cases they speak clearly and there are subtitles (although there are exceptions)

After viewing all the videos in the module, the idea is to have a laboratory, but in this course it is not there, instead of it there will be 1 page from the book in the modules, there will be questions from this page in the tests for the module, the questions are simple, just glance text.

Actually, the final exam also does not shine with complexity and in fact is a test of knowledge on the miracle book from the laboratory.

On the other hand, this is an introductory course, we will not want miracles from it.

At the end of the course, you will at best understand why you had to invent this Data Science at all . Practical skills will not give any.

But then you will receive a badge that will proudly confirm what you are saying: "Its value is approximately the same as that of "Bunnies" who instead of grades in my childhood put in a notebook first graders

I will write about others even more concisely:

Data Science Methodology - Oddly enough, this course turned out to be difficult for me, and primarily because of the language barrier, if the specifics of English for IT are familiar to the eye, then the more scientific specificity of the language caused difficulties. By itself, the course of real practice in essence does not give, but describes the basic concepts, tells how Data Scientist should think about (I will call him that because the “data scientist” doesn’t sound so cool)

Unlike the previous course, this already has some semblance of laboratory, you will download a pdf notebook and if you want, even answer the questions put there (which is not necessary). You can’t do a couple of labs, because “you don’t know how to”, the authors suggest you if you want to come back to them later (I didn’t want to).

Data Science Hands- on with Open Source Tools - It would be strange if IBM as part of its courses did not advance its development, this course will introduce you to their datascientistworkbench.com toolkit . The stuff is free, it is hanging in the cloud, one minus the modules are not initialized very quickly. As part of the course, you will be taught to use apparently the basic open-source tools that are used for data processing (or maybe not, I'm a layman in this matter, so I will believe IBM). In addition to the introductory part, the main focus will be on the following applications: Jupyter Notebooks, Zeppelin Notebooks, RStudio IDE, Seahorse. Once again, everything hangs in the cloud does not need to put anything.

In contrast to the previous courses, this one, although it already offers a small practice, within the laboratory, it will be possible at least a little to indulge in tools, but the level of tasks is for quite beginners. In principle, it will be necessary, just to see how everything works.

R 101 - here we will learn in more detail the basics of the R language and their RStudio version. The laboratories are already acquiring at least some meaning, in some places it will even be necessary to tweak their code a little bit in order to get the numbers needed for answers in tests (if the memory did not let me down). But again, the course is for completely newbies, a bit more complicated than “Hello world”, so don’t harbor any illusions in programming in R, you can hardly learn it right away.

So, at the end of all the courses you receive certificates and the promised badge and then you may have problems.

Part 4. Certificates, badges and problems with closing the course.

Not that free cheese was only in a mousetrap, but we obviously will not look at the gift horse in the mouth.

I think the attentive reader has already guessed that the value of certificates and badges tends to zero.

The certificate is the same as elsewhere, you can share it by link, you can print it and paste it in a frame.

The vast majority of certificates and badges do not require verification, which means anyone can get them.

Now about the badges. Badges are posted on the partner site https://www.youracclaim.com . (it is necessary to create 1 more profile), there you can publicly display all your achievements and then share the link to the profile with everyone at once, for example in the social. networks or resumes.

Problems. Imagine you bravely passed all the courses to the second step, received all the certificates, but did not give you a badge. Do not be discouraged if you really need it, you can get a job in retail :) . We assume that the crossed out way does not suit us and begin to understand what the matter is.

If there is no badge, the first thing you should pay attention to is the Progress tab. The program will issue you a certificate as soon as you pass the threshold level (usually 70%), but the badge is more complicated. Be sure to ensure that you answer all the questions in the tests (clicked Final Check where required). If there is at least one unread question in one of the courses, the curriculum is not completely closed to you.

So, you went over the “progress” contribution for all the courses and were convinced that “know the problems” is written everywhere, but there is still no badge. Then begins shamanism, I recommend to open each course again and click on the “Courseware” button. If the text is different in meaning from:

“You have recently received your certificate and badge. If you’re done with that, choose another section on the left. ”

or from:

“You were the most recently in Download your completion certificate. If you’ve done that, choose

that is, it makes sense to go to those points where he advises. Be sure to recommend requesting a certificate at the bottom of the “progress” tab using a link like “Download your completion certificate” and poking it in there, I noticed that when you request a certificate on the progress tab from the top, it probably does not fix the fact that the training is completely over.

So, we have disassembled the start Data Science curriculum on the Cognitive class site, for those who are already tired of the big text I suggest to go to the conclusion at the end of the article, for the rest of the bonus - a brief description of several more courses.

Part 5. The “Data” knight or something else about Data Science

Since the first curriculum was completed in a day, and there was no increase in special knowledge, it was logical to go further, especially the curriculum developers themselves advised to switch to Data Science for Business , well, I also decided to look towards the course Statistics 101 , and I will start .

Statistics 101 - It seems that the guys were “blown away” to this course, because after a certain point no subtitles were made to the video, of course there is an automatic translation of Youtube, but this is not very convenient for me. With my bad English course it seems difficult, at least if you were not given statistics in the university before, it is difficult to expect that you will get insight in the “basurmanian” language. Nevertheless, the course is simple and something useful reports (standard deviation, median, variance, etc.). Maybe it makes sense to look at it before the courses on Data Science, and maybe not, it's up to you.

It is important to note that for this course you need to register to download the trial version of SPSS Statistics. The trial is only 14 days, so at least formally, the course time is not limited, it is better not to delay it at all. The program itself is expensive, and the course is largely tied to it, so ultimately I did not like the course =)

Data Science for Business

Consists of:

1. Data Privacy Fundamentals
2. Digital Analytics & Regression
3. Predictive Modeling Fundamentals

For the first two courses there are no subtitles for video (sort of).

Briefly about each:

Data Privacy Fundamentals - a course on the example of Canadian law shows how important it is to comply with information security. In addition to the text and video, there will be one exercise in the course, where with the help of blanks on R, we will be told how easy to crack unreliable passwords. Well, the exam will have to "hack" the password for poor guy Justin (you can not "hack", but simply "turn on your head")

Digital Analytics & Regression - the course finally gives at least a little bit of adequate practice and demonstration of the analysis of "small" data on R. Not that anything, but still useful.

Predictive Modeling Fundamentals I is a terrible course, a guy often sizzles into a microphone in a video and speaks as if he has a mouth ... a lollipop, and I don’t know a glitch or not, but on youtube the video is not uploaded, and the subtitles cannot be turned on in the player , so that they come out as normal subtitles (it turns out only from the side) as a result, watching the video turns into torture.

The course is sharpened by another creation from IBM - SPSS Modeler, so you need to download the trial again (this time for 30 days)

Unlike past courses, the material for the laboratory is not normally prepared, and if you make a mistake somewhere, you have to look for the solution yourself, check with nothing, there is no control file.

As part of the course, they analyze the problem about Titanic

I didn’t master it completely, I gave up after the second module, just skimmed the video, got into Wikipedia in difficult places, answered the tests + exam and successfully passed this course (which again speaks of the low value of certificates).

Conclusion:

Assuming that at the time of the “fork” not everyone read Part 5, so I’ll share my impressions into two parts.

Only Data Science Fundamentals completed:

Well, on the whole it is enough to understand in a very general way what Data Science is. No training is required, nor mat. analysis can neither be owned by statistics, nor by programming, the main thing is “to go in aglitski”.

I think it’s so obvious that in one day you don’t really learn anything and you shouldn’t expect a salary of 1,500,000 million rubles (I hope you haven’t had time to open Hunter and create a resume?)

In theory, this course should develop your interest in the subject and not scare, in principle, the developers succeeded.

Finished by Data Science Fundamentals + Data Science for Business + Statistics 101:

It destroys all hopes, because the truly sensible practice never got caught, and Data Science for Business + Statistics 101 courses are somewhat worse in terms of quality than Data Science Fundamentals, and they also require installation of trial versions of programs from IBM.

The examples in the problems are not anyhow and are cut off in many ways from reality.

Probably, having gone through all this, you might make a conclusion for yourself - Data Science is yours or not, were you terribly bored, or were you in awe of sorcery over the data?

To summarize : The courses presented in favor resemble a situation as if you couldn’t drive a car, put a normal car with an automatic transmission behind the wheel, show you gas and brake, how to start a car and fill up with gasoline, how to turn on headlights and wipers , well, in the end, under control, they would let us drive a couple of kilometers along a country road. On the one hand, you will not become a driver after this, on the other hand, if you save yourself from a maniac with a chainsaw, perhaps this knowledge will save your life. Exactly also with these courses.

In any case, everyone who has spent time learning from the Cognitive class program, I advise you not to stop there. In the end, even they have a lot of interesting things there (Big Data, Hadoop, Scala, etc.)

If colleagues recommend in the comments, the truly suitable free resources will post them in the update to the article.

Thank you for your attention, all a good week!