Transcription
Frequently, frequently, a question is asked: how to learn to be a product analyst? Andrey is a product analyst, you should know how to learn to be a product analyst. Often, people come with some unclear courses and ask, "Is this normal? Is this normal?" And so, guys, to stop all your questions in the chat, and especially in private messages, I'm tired of answering this. I will explain it clearly, that is, all that I know for sure are good ways to become a quality product analyst. Everything that a product analyst can take, it stands on 5 turtles. The first is SQL. You need to pass SQL. SQL is a language for working with databases. If you are already confused here and don't know what SQL is, look, I have two videos about SQL. Accordingly, the second is Python. Python is also necessary for data analysis. If we extract data using SQL, somehow preprocess the data with Python, and analyze it. The third is statistics. That is, you need to understand simple statistical laws. You don't need to be a statistician. You need to have statistics in the necessary amount to conduct A/B tests and to analyze some simple correlations or dependencies. The fourth is visualization tools. Any product analytics does not live without dashboards, without some visualizations, graphs, and everything else. That is, this is not the most difficult part. And the fifth part is product understanding. Product understanding includes, accordingly, product metrics, product hypotheses, how all this functions in Agile. We are all on the site. Accordingly, where to learn all these five turtles? First, SQL. I have an excellent video about what is included in SQL. I have a video with task breakdowns in SQL. I talked about this in detail. Let's go through it quickly. For theory, there is this article on Habr. For practice, there is SQL Zoo. It's excellent, just take it right away, there is practice plus theory. For practice, there is Interview Query. For practice, there is StrataScratch. That is, take these resources all together and learn, learn, learn, learn, learn, learn, practice, practice, practice. Often, the tasks that are asked in SQL are not very complex. SQL has a certain ceiling of complexity. That is, when you pass it, it's not like Python. Python can be mastered for a sufficiently long time, and new methods can be found because with its help, you can do a lot. With SQL, you can also do a lot, but mainly 90 percent you will learn in a month or two. Second, Python. What is needed for learning Python? There are good free courses on Stepik. You can watch them. On the one hand, on the other hand, I always recommend, I like the course "Mathematics and Python for Data Analysis." It was made by Yandex, it was made by M41, a respected result, another respected university. And there they give quite a lot at once, they stuff knowledge in math, statistics, and Python. Everything is there at once. And if you don't understand something, you then google it separately, go through it somehow. But in principle, for work, you can say that this is enough. But you can, for example, take this course and then additionally add some tasks for yourself again in Python. This will bring understanding. You understand what the deal is. Again, it's not like Python is a rocket science. You just need to implement some statistical functions, some simple calculations, some computations using Python. That is, all this for analysts is often enough. In principle, a few libraries are enough. That is, on the one hand, there is SciPy, and there is NumPy, there is Pandas, there is Matplotlib. NumPy is for working with numbers, with numerical data, and everything else. Pandas is for working with tables, you can say. SciPy is for various calculations, there are excellent modules, for example, a module with statistics where statistical calculations are performed, and Matplotlib is for building graphs. Accordingly, these libraries you master, read the documentation, learn to apply them, practice, and that's it, congratulations, you are a specialist. Well, almost. Next, statistics. Where to learn statistics? That's the biggest hurdle, it often happens in work. That is, you've mastered the tool, but not statistics. Well, I advise and recommend the YouTube channel "StatQuest." Some statistical concepts that were unclear to me at university, they are explained quite well there, and you can immediately apply it. That is, there is a little bit about machine learning, immediately about product analytics. If that's not enough, you can look at our courses. On the one hand, on the other hand, there are simple statistical concepts. I highly recommend it. If this is not enough for you, again, in the previous course "Mathematics and Python for Data Analysis," there is a large chapter on A/B tests, experiments, statistics, and everything else. If you still don't have enough, many praise Karpov's courses. That is, there are courses on statistics by Karpov. I think they are already in open access. In general, I haven't taken them, I don't know. But you know, everyone I meet, psychologists say Karpov, Karpov, Karpov. So, you can take them. You will mainly need 90 percent of the reality, you will need to understand how A/B tests are conducted. That is, what are confidence intervals, how are these confidence intervals constructed, how to calculate, for example, the difference in confidence intervals. Well, in general, how to conduct experiments, and in principle, everything that is related to experiments. If you know this, if you come to an interview and say, "Oh, I'm well-versed in A/B tests, I can read cohort analysis, I can interpret the results correctly," you will be hired with open arms. Accordingly, if you grow further, become a senior, senior, lead, and lead to heaven, then of course you need to learn further, learn deeper. That is, you can learn endlessly here. But if you are not aiming for this yet, then just take StatQuest, take Karpov's courses, take this course "Mathematics and Python for Data Analysis," and you will have happiness and help. Everything is excellent. Next, visualization tools. Yes, such a thing that you don't necessarily need to learn separately. That is, you just need to dig a little. For example, Tableau. Now, job postings often write, for example, "we build reports on our dashboards." I don't know. I haven't worked with dashboards, but I can guess that everything is the same in principle. Visualization tools are similar to each other. And more importantly, they are surprisingly similar. In my experience, they overlap with SQL. That is, how I work with SQL, the same thing I see. I don't know, some neurons are connected. They are similar. The same is Tableau. Just download it, look at some test dashboards that are available. You can download some test datasets if you don't know where to get datasets to analyze and play with them, look in the sandbox. There is such a wonderful site, Kaggle. Competitions on description and data analysis are held there, and there are many such cool, simple datasets. For example, there is a dataset on the probability of death on the Titanic. That is, a list of passengers, and then whether they died or not. You can build some machine learning models with it, or you can just see that all the men died. And try, try, try. That is, again, there is no rocket science here. You just need to get some experience. And then you will be able to work normally. By the way, what's better than learning Tableau, Data Studio, some dashboard, what else, Power BI? Better than all of this, look at the book "The Visual Display of Quantitative Information." It's an excellent book about how to approach visualization. If you learn to do this, you will be priceless because, well, I am also learning this and actively progressing in it, because the same graph, well, it seems normal, it describes, can I look at their tables? Yes, correct results. But how are visualizations built there? How are examples chosen? When are which visualizations better? But this is perceived separately. I think it was in some Telegram channel, then he wrote a book. Well, in general, check it out for sure. And so, in general, you've dug into visualization, approached statistics seriously, approached Python seriously, approached SQL seriously. Well, and the simplest part is left: product understanding. Product understanding. There are no good courses for product understanding. But there is GoPractice. But GoPractice is a bit overpriced for getting this product understanding. That is, it costs 40 thousand. This is growth when you don't work in IT and want to enter. Accordingly, you can watch my videos about product hypotheses, about A/B tests, about Scrum, about Agile. On the one hand, on the other hand, there are many materials in open access. Again, there is the same GoPractice blog, where they write about A/B tests, about products, about game economics, and about everything. There are lectures from Yandex. It's absolutely wonderful. Yandex had a school, as far as I remember, a school for managers, and they have many lectures in principle in open access. Just absorb them all, and you will already have some understanding. That is, in fact, you need to understand metrics, ways of product development, this product lifecycle, who is responsible for what, in order to be a product analyst. So that you don't get lost at work. Just what is happening here is unclear. Someone is doing stand-ups, retrospectives. Someone is generating product hypotheses. This is unclear. Accordingly, you can just get this from YouTube videos, and nothing will be wrong with you. There are no special courses on product metrics. That is, just watch my videos, all of which have something about products in one way or another. Watch these lectures, read various blogs. That is, go to VC.ru. Just this is general education. If you have this general education in conjunction with the hard skills from my previous points, you will be an excellent product analyst. So, we have covered, in fact, 5 turtles. If you know any other good courses, resources, or advice, or you will say now, "No, you are wrong, this is not how it should be learned, it should be learned completely differently," then definitely go to the comments, join the chat. Again, don't forget to watch this video, subscribe, hit the bell. Instead of asking me questions in private messages on how to learn to be a product analyst, better hit the bell, because I consider every, every, every bell. Thank you, and good luck with your studies. It will be difficult, you will learn a lot, but you will succeed. Bye.