Compare and contrast is a word that is certainly not unfamiliar to all of us, for example, if we buy something we will compare and contrast, in fact, there are comparisons everywhere in life.

For example, Xiaofang has always been excellent, but the last exam played out of order, only 40 points in math, the class teacher found Xiaofang talk, asked her, "What's the matter with you lately, last time you got 80 points in math, the top ten in the class, this time how to test so bad? Look at your deskmates, they all scored 73 this time."

From this short story, we can see that there are generally two ways to compare, horizontal comparison and vertical comparison. Horizontal comparison is also known as comparing with the same kind, for example, the class teacher compares Xiaofang's score with her classmate's score. Vertical comparison refers to the same type of comparison at different times, for example, the class teacher takes Xiaofang's grades this time and last time's grades to do a comparison.

2. Segmentation Thinking

Segmentation thinking is something that many people may not understand at first glance, but in fact many small things in life reflect segmentation thinking. For example, the human body is composed of nine systems, the system is composed of organs, organs are composed of tissues, cells and constitute the organization, the layers of subdivision.

Take the example just now, or our Xiaofang students, or just that exam, the class teacher let Xiaofang to do a summary of their total results of this exam, Xiaofang took the report card carefully, found that the total results of this is not very good, but take a closer look, found that in addition to the math scores of only 40 points, the results of the other subjects are at the top of the list, the math scores pulled Xiaofang's overall grade.

Here we are summarizing the attribution by breaking down the overall test scores into specific subjects. In the work of data analysis, the latitude of segmentation mainly includes time, region, channel, product, employee, customer and so on. DuPont analysis, McKinsey's MECE analysis method essentially belongs to the segmentation thinking.

3. Traceability thinking

The first two thinking can correspond to part of the data analysis requirements, but if there are some data can not use the first two thinking to deal with how to do?

Then we can use another kind of traceability thinking. As the saying goes, to trace the roots, many times we want to know the logical reasons behind things, the best way may be to explore the reasons why things happen to help us analyze.

Continuing to take Xiaofang as an example, she came home from school and gave her report card to her mother, who knew the general situation of Xiaofang's test through the comparison and breakdown method of thinking, and also knew that she had failed in math. But Xiaofang's math has always been a strong point, the mother still can not understand why there will be a problem here, so the mother to find Xiaofang to talk about the detailed understanding of the situation during the test, only to find out because Xiaofang the day of the math test at noon ate a bad stomach, the afternoon of the math test happened to have a seizure, the pain is intolerable, so that a lot of would have been able to do the questions are wrong. Mom also understood Xiaofang and apologized, and will pay more attention to Xiaofang's diet.

In the example above, Xiaofang's mom was unable to analyze why things were happening from the surface data, and used retrospective thinking to find the real cause. If data analysts can also utilize retrospective thinking in their work, then their sensitivity to data and understanding of the business can be gradually deepened.

4. Relevant thinking

The above kinds of thinking are more commonly used thinking, the following we will talk about relevant thinking, which is also the core thinking ability of data analysis.

Many people may know the famous story of beer and diapers, which is a classic example of correlation analysis in the industry. The story is set in the 1990s in the United States Walmart supermarket, when Walmart has the world's largest data warehouse system, in order to be able to accurately understand the customer's purchasing habits in its stores, Walmart shopping behavior of its customers shopping basket analysis, want to know what customers often together to buy goods.

Wal-Mart's data warehouse centralizes detailed raw transaction data from its stores. Based on this raw transaction data, Walmart used data mining methods to analyze and mine this data. One unexpected discovery was that the most purchased item along with diapers was beer.

After a large number of practical investigation and analysis, revealing a hidden in the "diapers and beer" after the Americans a behavioral pattern: in the United States, some young fathers after work often go to the supermarket to buy baby diapers, and 30% to 40% of them at the same time also for their own to buy some beer. The reason for this phenomenon is that American wives often urge their husbands to buy diapers for their children after work, and the husbands bring back their favorite beer along with the diapers.

If data analysts are skilled and flexible in applying relevant analysis to their work, they can progress from just knowing what the result of data analysis is to knowing why that result is presented.

5. Hypothetical thinking

The previous thinking mode is based on a large amount of known data can be analyzed when the argument, so if we do not have enough data volume or evidence to verify the matter, what should we do? This kind of time we can use our hypothetical thinking. First of all, the bold assumptions, and then careful evidence, and finally to find ways to verify whether the assumptions are valid.

For example, Xiaofang wanted to eat lychee, so she went downstairs to buy it, and there was such a conversation with the lychee seller's aunt:

Xiaofang: "Auntie, is this lychee sweet?"

Auntie: "It's sweet, I've got some cut ones here, you can try it first."

Xiaofang: "OK, I'll try one then."

Xiaofang brought a lychee and tasted it, "Well, it's good, it's really quite sweet, weigh me two pounds."

The above seemingly simple little story actually hides a simple hypothesis test. First, Xiaofang makes the hypothesis that lychees are sweet; second, a random sample is taken; then, it is tested to see if it is sweet; and finally, a judgment is made to confirm that the lychees are really sweet, and so they are purchased.

In data analysis, the technical term for hypothetical thinking is called hypothesis testing, which generally consists of four steps, i.e., formulating a hypothesis, taking a sample, testing the hypothesis, and making a judgment. Data analysts can take full advantage of this mode of thinking.

6. Reverse Thinking

The term "reverse thinking" must not be unfamiliar to anyone, and is often mentioned in the speeches of many famous entrepreneurs, who advocate breaking out of the conventional mode of thinking and thinking in the opposite direction.

Now we invite Mr. Fang to come on stage again.

On one occasion, Xiaofang went to buy chili peppers and had another conversation with her aunt.

Xiaofang: "Auntie, how much do you charge for a catty of this pepper?"

Auntie: "One dollar five."

Xiaofang picked 3 and put them on the scale: "Auntie, help me weigh them."

Auntie: "One and a half catty, two dollars and twenty cents."

Xiaofang removed the biggest chili pepper: "I don't need that many to make soup."

The stall owner: "One catty, two taels, one yuan 60 cents."

Xiaofang picked up the biggest pepper she had just removed, paid 60 cents, smiled, and said goodbye to her aunt.

You see, using reverse thinking may sometimes have an unexpected effect.

7. Deductive Thinking

Deductive thinking may not be so well understood compared with the previous ways of thinking.

The direction of deductive thinking is from the general to the individual, we should remember this point, we will mention later. That is to say, the premise of deduction is general abstract knowledge, while the conclusion is individualized concrete knowledge. The main form of deduction is a syllogism consisting of a major premise, a minor premise, and a conclusion.

Take, for example, a general knowledge in physics.

Major premise: Metals conduct electricity.

Minor premise: Silver and iron are metals.

Conclusion: silver conducts electricity.

From this example, it can be seen that the major premise is a known general principle (that metals can conduct electricity), the minor premise is the special occasion studied (that iron is a metal), and the conclusion is the new knowledge derived from grouping the special occasion under the general principle (that silver can conduct electricity).

8. Inductive Thinking

Inductive thinking is the opposite of deductive; the process of induction is from the individual to the general.

Or take the example that metals can conduct electricity.

Premises: gold conducts electricity, silver conducts electricity, copper conducts electricity, and aluminum conducts electricity.

Conclusion: metals can conduct electricity.

The process of data analysis is often the first exposure to individual things, and then inductive summary, and then general, and then deductive reasoning, from the general to the individual, and so on and so forth, and continue to accumulate experience.

Summary

This article summarizes 8 kinds of thinking in data analysis, which are comparison, segmentation, traceability, correlation, hypothesis, inverse, deduction, and induction. As a data analyst, if you can fully utilize these thoughts in your work, it is a great enhancement to your personal ability, and you can create more personal value in your work.