# 7 classic data analysis thinking! (2) - Zhihu

2020-12-08 08:50:22

Today, Lao Li continues to explain the classic thinking model of data analysis , The first part introduces goal thinking 、 Hypothetical thinking 、 Traceability thinking 、 Converse thinking 4 Thoughts , Today I will continue to introduce Structural thinking 、 Deductive reasoning thinking 、 Summarize thinking and related thinking .

If you miss the last one, you can review this article ：

Li Qifang ： Data analysis must know ： A detailed explanation of the seven classic thinking ！zhuanlan.zhihu.com

## 1、 Structural thinking

Most of the time, people don't have ideas , I don't know where to start , This is the lack of structured thinking

Let's take a look at the following example , Let's see if you have structured thinking ：

An offline retailer's sales of a product have dropped recently , Let's find out what caused the decrease in sales .

Let's take a look at the analysis ideas of a and B ？

nail ：

• First of all, analyze from the time dimension , Look at the drop in sales. It's a sudden drop , Or a sustained decline ;
• Then take the store as the dimension , Let's see if the drop is due to geographical location ;
• besides , And compare horizontal competitors , You can ask some salesmen what they have ;
• by the way , There are also activities , It's likely that sales are down due to the event .

It's a mess, isn't it ？

This is because when we think about problems , Get used to point-to-point , A little thought is a little bit

That is to say Shooting at random , Maybe you can use experience to find out why

But most of the time , It's hard for you to find the reason why it's completely exhausted , That's why your data analysis always has no idea

B ：

• The problem we are going to analyze is the decrease in sales , Generally speaking, there are internal and external reasons
• Inside, it's something within itself that causes the decline , External causes are force majeure factors beyond our control
• We can refer to 5w2h Several factors in ,when、why、who、how etc.
• External factors include market competition 、 Market capacity 、 Policy, etc

Knowing these key factors , We'll go on with the disassembly , To find out all the possible causes

This analysis is not a lot of clear feeling ？

How does structured thinking approach deal with this problem ？ In the face of such a problem , The first thing structured thinking does is not to clean up the data immediately .

It's based on an understanding of the business , First draw a mind map for data analysis , Its function is equivalent to that you come to a strange city and take out Baidu map to query the route map from transportation to hotel .

This mind map is a map of the route to your destination .

in fact , Structured thinking is the famous one put forward by McKinsey “ Pyramid thinking ”, As shown in the figure below is a typical structure ：

Whether it's as an expression 、 Or the receiver of information , We have to build a framework that conforms to the pyramid structure , And then logically 、 The key contents are described in order

And about the pyramid structure , The key to my understanding is “ The main - important - secondary ”

Among them “ The main ” It's to clarify the central idea , This book puts forward 4 Requirements ：“ Conclusion first 、 Upper system and lower system 、 Classification and grouping 、 Logical progression ”, These are the four principles of the pyramid

Among them “ important ” It's when you build the pyramid , Be sure to follow the rules first and then the second 、 First overall, then detailed 、 Conclusion before cause 、 The principle of "result first, process first" is used to arrange the content

final “ secondary ” It's about putting irrelevant 、 Poor logic 、 Low correlation factors and content were screened out

More specifically, it's ：

• Conclusion first ： The central thinking should be put at the front
• Upper system and lower system ： The upper layer must summarize the content of the next layer
• Classification and grouping ： The ideas of each group should belong to the same logical scope
• Logical advancement ： The order of each group should follow a certain logical relationship

## 2、 Induction and deduction

First of all, what is induction and reasoning ？ Let me just give you an example ：

• inductive ： Trees can burn 、 Paper can burn 、 Chopsticks can burn , So wood can burn
• Reasoning ： Wood can burn , Chopsticks belong to wood products , So chopsticks can burn .

Obviously , Induction is based on individual attributes , Look for commonalities between factors , To sum up a general feature

And deduction is the opposite , It's from the general whole , Looking for logic between things , To get the characteristics of an individual

In a real business analysis scenario , We will imperceptibly use deductive and inductive thinking , For example, deductive method , Our most common is syllogism ： The big premise 、 The premise and conclusion .

But deductive method should pay attention to avoid a big mistake ： such as “ Recently, the company's profit margin has declined , Because the cost is too high , So we're going to lower everyone's pay .”

First of all, this argument is based on the deductive syllogism , The logical relationship between each paragraph is correct , Profit margins do have to do with high costs , And the cost also includes the cost of human salary , It seems as if the logic is tight , But if that's true , Maybe every company will use this reason to lay off workers and cut wages

What's the problem ？

Clearly, the relationship between each round is logical , The problem is between the big premise and the small premise Whether the argument is really persuasive

For example, whether the company's profit margin is just because of the high cost ？ This is the argument for the grand premise

If the cost is too high, it can only reduce everyone's salary ？ This is the argument for the minor premise

Obviously , The argumentation process of these two premises is not rigorous , So there's a logic barrier .

Then induction is simpler , Induction is based on the result , Find out why , By observing and comparing 、 analysis , A way to find causality between things

alike , There is also a mistake in induction ： Black swan event .

Every morning the farmer 7 Feed the chickens in the chicken farm on time , Over time, the turkeys came to the conclusion that farmers 7 They'll feed the chickens , But it's a knife that's waiting for the turkeys on Christmas day .

This is a fatal mistake in induction , That is to say, to generalize , We can't stop the black swan

3、 Related thinking

In the age of big data , The core is related thinking , This kind of thinking is based on correlation analysis .

The story of beer and diapers , It's a classic case of correlation analysis .

The story comes from 20 century 90 In the Wal Mart supermarket in the s , At that time, Wal Mart had the world's largest data warehouse system

In order to be able to accurately understand customers' buying habits in their stores , Wal Mart conducts a basket analysis of its customers' shopping behavior , Want to know what goods customers often buy together

Wal Mart uses data mining methods to analyze and mine these data , An unexpected discovery is ：

Beer is the most popular product to buy with diapers

After a lot of practical investigation and Analysis , Reveals a hidden in 「 Diapers and beer 」 A pattern of American behavior behind it ：

In the U.S. , Some young fathers often go to the supermarket to buy baby diapers after work , And some of them 30%～40% At the same time, they buy some beer for themselves

The reason for this phenomenon is that ： American wives often tell their husbands to buy diapers for their children after work , After buying diapers, the husbands brought back their favorite beer .

A simple example , Generally speaking, what women buy in supermarkets is cosmetics 、 clothing 、 Vegetables and so on , Most of the things men buy in supermarkets are daily necessities , So there will be women's and men's counters in supermarkets , Through simple customer grouping to achieve commodity classification .

https://chowdera.com/2020/12/20201208085009994m.html