Why do you have no idea when doing data analysis?

Many students complain ：“ No idea when doing data analysis ！” actually , There are many reasons for this result . Check the system today .

There are some cases , It's the data analyst's own problem , There are three common ：

problem 1 ： Take a hammer to find a nail

mathematics 、 statistical 、 Operations research has many methods , And reading itself can make people feel full . therefore , There are some students who are fascinated by reading , Start looking for nails with a hammer . For example, today we see the statistical normal distribution , That's amazing , So everyone looks like a normal . See the chapter of regression analysis tomorrow , Everyone wants to come back ……

Doing so will cause trouble . For example, some students calculate the benefits of activities , Return the non activity funds to the total performance , Then look at R The square value says that the activity has no effect . The result is naturally sprayed by the business .

and , I didn't really understand the book , If you really understand , At least distinguish ：

• Is it sampling statistics or overall statistics

• Is it a prediction problem or a classification problem

• Yes, there are labels but no labels

• Is there any inherent logic in the data at hand

In depth business scenarios , Just know which method is suitable for the book . And all kinds of Book Methods , There are fixed application scenarios .

problem 2 ： Take your shoes and cover your feet

This is the brother of the last kind of problem , It's all nerd behavior , But the book from 《 statistical 》 Instead of 《 Management 》：

Because the book has 4P, So draw four first P The frame of

Because the book has PEST, So draw four frames first

Because the book has RFM, So count first RFM

so what ？…… Then I got dizzy , I don't know what to do , Then rated as ：“ What are you doing ……”

The solution is the same as the previous problem , First understand the business scenario , Find the real problem , Reorganization method . Instead of taking a condom first , Cover everything you see . Data analysis serves the business , How much does the business know about the problem , Is the starting point of the analysis （ Here's the picture ）.

problem 3： Take apart everything

It's also very common , No matter what the problem is , Pull a bunch of cross tables first .

For example, analysis DAU, Just put DAU And gender 、 One pass intersection of age and other dimensions

For example, analysis GMV, Just put GMV And gender 、 One pass intersection of age and other dimensions

It's called ： The soul of data analysis is comparison , The core is disassembly

The result is ： Missing logic in , Without assumptions , The more comparisons , The more confused the thinking is . Often do the embarrassing thing of comparing apples with rhinoceros . And this aimless intersection , Often lead the business thinking astray . The business department will catch you , Let you explain one sentence at a time ： Why are there differences here 5%, There are differences 3%, Finally, the more confused the idea is ……

So comparison can be done , But first list the assumptions , Label it , Make apple better than apple , To find out .

There are some cases , It's not necessarily a matter of data , The data is just a back pot , There are four common ones ：

problem 4： No business objectives

such as ：

Do indicator monitoring , What are the index assessment requirements ？ I do not know!

Do activity analysis , What indicators should the activity improve ？ I do not know!

Do product analysis , What is the purpose of product revision ？ I do not know!

Then I don't know how to analyze ……

In this case , I really don't know how to analyze . It's like archery , You have to have a target first , Just know whether the shot is accurate or not . Not even a target , Close your eyes “ Walk you ！” Shoot indiscriminately , Then let's analyze whether this random shooting has the effect of changing the day , Ah bah ！ Analyze a fart .

Of course , Most of this problem is caused by business . But remind the students who do analysis , Be sure to ask about your goals first . And actively prompt the business department ： The goal is not clear , The analysis is naturally unclear . Otherwise, it's easy to be thrown out of the pot here . Many business units , Set no goals for yourself , Then force the data analyst to write ：“ This indiscriminate shooting has greatly improved the company's performance ！” Then when I was scolded by my boss , Just say “ This is all written by data analysis , I'm innocent ”……

There are a lot of ways to set goals , Don't say you don't know how to decide （ Here's the picture ）.

problem 5： Mixed goals , Deceive oneself and others

This problem , yes “ No target ” The opposite of , Just a little work done by the business department , The result was earth shaking . For example, they put in a 10 Yuan coupons , Then start blowing ： This piece of 10 Yuan coupon , It can be pulled up GMV, It can wake up old users , The middle can promote new users . Anyway, the effect is great , Then ask the data analysis to analyze how much each effect is , Also have to give advice that can be landed ……

Many students are confused , What are these ！ How do I analyze it ！ I don't know what to do , Because it's just fooling around . Each type of business practice , Have a fixed form . Not so much “ All over the world ” How to do it . So I want to form ideas , We have to understand the common routines and fixed forms of business , In this way, in the face of such nonsense, we can distinguish the situation .

problem 6： Poor business means

The most common is ： User portrait 、 Loss prediction 、 Recommend these items , A lot of data , Business has no tricks .

You worked hard to predict the loss probability of different users , The result? ？ The business side will send a full text message recall …… SMS yeah, The full volume response rate is less than 1%, What's the difference between making a model or not . And then in the end , Make complaints about business ：“ Your analysis is useless ”

So is product recommendation , Many companies can't get a few powerful products . Recommend... For recommendation , A whole bunch of uncompetitive junk goods , Then he ran over and asked ：“ Why analysis is useless .”

This kind of questioning , It will make the students who do the data fall into deep self doubt “ Am I thinking wrong ”. However, there is no need to doubt , It's not a question of thinking , But the lack of business means leads to the problem of being unable to land .

Want to break this game ： Basic analysis must be done well , For goods 、 user 、 Have a basic understanding of business means , I know how much capacity the company has at present , It's easy to identify ： What is wrong with me , Or these guys, just these two ？

problem 7： Missing iteration , No accumulation

Good data analysis model , It's iterative , It didn't fall from the world . Define a goal , Run multiple tests , Find out the capacity limit of each business means 、 Lower limit , So you can see which method works , To discover the inner logic , To accumulate analytical experience , This is the right way .

But some enterprises just like to go astray , such as ：

1、 Make business plan “ Both XX, And XX, still more XX, synergy XX, Together XX” There are a lot of goals , It's not clear which direction to measure

2、 Let analysis every day 、 analysis 、 reanalysis , Just don't do the test

3、 Let analysis every day 、 analysis 、 reanalysis , Analysis of finished , The business uses a completely different set of ideas to test

4、 Fail to reach the goal , Just change the goal , present a false appearance of peace and prosperity .

To do so , Like a headless fly . There is no need to form effective experience accumulation , In the end, there will be no harvest . However, the students who are deeply involved , I just feel my brain buzzing , Like to doubt ： Is it that I don't have a clear idea …… This is really not , This is a standard random work .

Summary

Data analysis should be closely combined with business , The same is true of analytical thinking ：

• Combined with specific business scenarios

• Have clear questions and goals

• Demonstrate logically

• Check the results through the test

• Accumulated experience through multiple rounds of testing

This is the right way to make your analysis ideas clearer and clearer .

Of course , Some companies just have a bad environment , As a result, students are always ignored in their work PUA“ You don't have a clear idea ”. Now , As long as you do your job well , Accumulate more practical experience on specific problems , There is a chance to change jobs and leave such brain disabled companies , Find a better job . So it's important to discuss the details , Don't think about the details , Theoretical or superficial , You'll make a fool of yourself as you did at the beginning of this article .

Some students will say ： Then I usually work , No one discussed the method , To do , I recommend you to join my knowledge planet . Knowledge planet on existing services , The following updates will continue ：

For entry-level students , Will be updated 《 Technical interview guide 》 Further improve the students' interview success rate .

For progressive students , Will be updated 《 Seven analysis scenarios 》 series . In addition to the existing 《 Business analysis 》 Outside , Will analyze users 、 Activity analysis 、 Commodity analysis 、 Channel analysis 、 Sales analysis 、 Product analysis and other topics , Update one every month . Improve students' ability to solve specific problems .

Knowledge of the planet Four highlights

1、 Discuss specific problems with Mr. Chen one-on-one

2、《 Ask for help 》 Course + One on one resume interview Counseling

3、300+ Data analysis topic sharing ,150+ Dry goods information

4、《 Nine methods of data analysis 》《 Business analysis 》 Video course system learning

Scan the code and join Mr. Chen's knowledge planet

Always ask questions

700+ My little companion is waiting for you

   

Click on “ Read the original ” Learn more about