Beware! Eight common misunderstandings in learning data analysis

Data analysis is a popular skill that attracts many learners. However, there are many misunderstandings in the learning process, which can lead to wasting time and energy. This article is carefully written by Mr. Chen, an experienced data director, and details the eight common misunderstandings when learning data analysis.

Is data analysis easy to learn? It is very easy to learn! However, many students fall into misunderstandings while learning, which not only wastes their efforts but also fails to solve the problem. Today, Teacher Chen will review the eight common misunderstandings. Students should correct them if they have any, and encourage others if they don't.

1. Unclear goals and greed for more

Why do you want to learn data analysis? Common answers:

1. I want to change my job to data analysis

2. I want to increase my salary

3. I am interested in data

4. I need to solve a specific problem

5. I think "big data" and "artificial intelligence" are very fashionable

If you look at each idea separately, it is correct. If you stick to each idea, you can achieve your goal. But the fear is: mixing these things together and achieving all at once. If you want to learn a skill that can solve the problem at hand, and find a job in a big company in the future, and the job is in line with your interests, and the salary income can increase several times, and quickly achieve all the above goals...

History has repeatedly taught us that the more words you have, the higher the probability of failure. Having too many goals will directly lead to learning failure. We often see an ambitious student buy 20 books at once, from "Statistics" to "Watermelon Book", from "Learn Python in 21 Days from 0 Basics" to "Who Says Novices Can't Do Data Analysis"... and spread them all over the table, but after a few months, he didn't learn anything.

In essence, data analysis is a typical cross-disciplinary knowledge, involving many subjects (as shown below). If the goal is not clear, it is very likely that you will be stuck in a certain branch and cannot extricate yourself.

2. Being limited to one corner and dwelling on a horn

In contrast to the previous situation, some students take a dead end approach.

"I am the man who wants to be the Excel King"

"I am the man who will become the SQL King"

"I am the man who will become the python king"

…

There is nothing wrong with delving into a technology. What is worrying is that people delve into technology not out of love for the technology, but because they think, “I will get a promotion and a raise after I become the master of Excel/SQL/Python…”

Um, clear answer: No

Firstly, it is too difficult to achieve "The King's Man" and it is simply impossible in a short period of time.

Secondly, the company does not pay according to software/tools/skills, but according to the position.

A job requires solving specific problems, which often requires a combination of multiple skills combined with reality. If you focus on one thing, you will often study hard for half a year and still not be able to solve any problems, let alone get a promotion or a raise.

3. Mixing job search and promotion

When setting learning goals, whether you want to improve your adaptability to your current position/industry or improve your competitiveness in the job market are two fundamental directional issues.

Improve the competitiveness of the talent market: meet the needs of future intended enterprises

Improve the adaptability of current positions: meet the needs of current enterprises

Unless the future job seeker has a similar industry status and business scenarios to the current company, what meets current needs is definitely not the same as what meets future needs. Not to mention cross-industry needs. The following are some common situations. Students can find their own situation and design a good learning route.

4. Job hopping across industries and insufficient training

This is a common question for students who are changing fields, especially when the gap is very large and past experience is of little help. Many students ask: "Teacher, I have learned Excel, SQL, and Python, what kind of job can I find?" At this time, I often ask back: "What do you mean by learned?!"

Many students just typed on the keyboard at online cases and book exercises, ran them and found that they could run the results, and then they thought they had learned it. The degree of training was too low, which led to forgetfulness during the written test and forgetfulness during the interview, and the result was definitely not good.

All technical operations require sufficient training. The best method is the "four sames" (as shown below)

5. Escape technology and bump into business

This is also a common problem for students who want to change their careers. Many students want to choose the business direction, not because they are good at business analysis, but because they think "I don't have the technical skills, so I'll just do business first..." Then when it comes to the interview, they don't even know what business the other party does, so it's natural that they fail in the end.

Note: Most of the so-called business analysts on the market are either "wild analysts" working alone in the business department, or cousins ​​who organize Excel spreadsheets for the business. Not only is their work entry-level and simple, but it will also be of no help if they change careers in the future. They still have to make up for their technical skills honestly.

Senior positions that truly do business analysis are often done by business personnel from strong business departments such as user operations, growth hacking, and channel promotion. In essence, it is a competition of business capabilities, not a pivot table in Excel.

I really want to develop in the business direction. When learning, what I need is to expand my knowledge, to have an in-depth understanding of the specific processes of a specific industry, and to conduct a detailed study of data collection methods and data forms. These knowledge can be supplemented by reading theoretical books such as "Growth Hacker" and "Lean Data Analysis", but a deeper understanding of the industry is required to a greater extent. Otherwise, once it comes to the specific operational level, you will be 100% at a loss (as shown below)

6. Job-hopping among peers, lack of depth

Many students who change jobs in the same industry will encounter this problem. The position they are interviewing for is obviously an ordinary analysis position, but they are confused by all kinds of difficult questions during the interview, and then they ask doubtfully:

“Does this kind of analysis require modeling?”

"Does this kind of analysis require theory?"

“Does this kind of analysis require a methodology?”

This situation is actually caused by recruitment involution. Everyone knows that they don’t need to use it, but the interviewer will still ask, otherwise how can we eliminate others? But if you really learn it, you won’t use it in your daily life, and you don’t know where to stop, what should you do?

When studying at this time, remember: breadth is greater than depth. First memorize the catalog of statistics, machine learning, and recommendation algorithms. Write down the common methods and basic ideas for each type of problem, and then practice cases when you have time, and then try to connect with reality when you have time.

This way, at least you won’t be completely stumped during the interview and can handle the situation. At the same time, when combined with your own experience, it is easy to say “these are the methods.”

Students who want to solve practical problems while working often encounter this problem. The most typical one is to open your mouth:

“What is the standard practice for user personas?”

“What is common practice for predictive modeling?”

“What is the scientific approach to cause analysis?”

If you ask him why he emphasizes "standard", "common" and "scientific", the answer is:

Leadership does not approve

Colleagues don't buy

Customer does not accept

This has nothing to do with "standard", "common" or "science". This is a question of how to deal with people. There is actually no book to refer to for this kind of question. If you want to read, you should read "Communication" or "Management" instead of "Advanced Predictive Modeling"...

There are many similar scenarios, such as

Issues that need to be resolved through communication: unified caliber, assessment objectives, and evaluation criteria

Need to be solved by management: lack of management standards, failure to implement standards

Need technical solutions: No digitization of workflow and lack of data collection

Need business solutions: lack of planning ideas, insufficient business capabilities, lack of experience accumulation

All these factors will make it difficult to perform data analysis in actual work. Moreover, these factors have nothing to do with data analysis. Therefore, if you want to solve these problems through learning, you cannot just focus on data analysis, but think from multiple angles: what should we do? Here is a simple criterion for judgment (as shown below):

8. Always think about the system and ignore the reality

Some students always think about practicing systematically during their daily study. This idea is not wrong, but it is easy to go wrong if you expect to practice all kinds of operations in a data set, or even a large wide table.

First, the data table structure is complex in actual work, and it is impossible to use a wide table for all purposes.

Secondly, in actual work, problems occur very scattered, and it is impossible to include all the problems in one table.

If you insist on finding a wide table to practice all your skills, you will find it difficult to find a 100% satisfactory data set, and even if you find a seemingly suitable data set for practice, you will still fail in actual combat after practicing. Because in actual combat, decomposing the problem is the first step.

If you really want to learn systematically, the first step is to break down clearly which are business problems, which are technical problems, which are data collection problems, and which are data extraction problems, and improve your own ability to distinguish. After that, you can improve your ability in each subject through monographs, so that you can handle problems with ease.

summary

The eight problems actually all stem from the initial one: unclear goals. Among all jobs, data analysis has the most theoretical support, and each subject is bottomless when you delve into it. Therefore, it gives people the illusion that as long as I read more books, I can conquer the world.

But in reality:

The working scenario of data analysis is very complex, mixed with theory, business, technology, goals, and human relations.

Job information for data analysis is very confusing, with all kinds of terms flying around, and there are many cases of selling dog meat under the guise of mutton.

These complex situations vary from person to person, and the difficulty level is not the same for people with different levels. Therefore, when you encounter a problem, don't rush to buy 20 books. First, sort out what you want. Highlight the key points and eliminate a few "and"s in your goals, and then it will be easy.