It's important to look at nature through phenomena.

In the early summer of the last century 90 's, the breeze was blowing and the cicada kept chirping. In the campus, each classroom has heard the rustle of Voices, a title of three students are in the middle of the tour. In the last row of the classroom, a comely girl with a ponytail passed a note to the boy next to it, saying, "I heard the cinema is playing Titanic today, we skipped class to see it!"

The boy saw the note, the face suffused with a faint blush, a face excited to return a note, "I heard that the people's Street video Hall today put the latest" ancient confused son "Ah, we go to see that bar. ”

That night, the man and a woman in the boy's insistence, or went to see the "Ancient confused". From the next day, the girl began to yell at him "brother". Two humans never again.

Many times, if the phenomenon can not see through the essence, it will certainly fall very miserably, chasing girls is so, playing the data is the same.

The different angles of reading the data from the priest

The same person, in the eyes of other different people's image is strange, it is difficult to do uniform. The same picture, the same data, are quietly in situ, different angles will bring different effects. For example, the same Taobao clothing, the effect of the monitor and receive the goods after the effect is difficult to maintain consistent, as shown in the body of the figure colleagues.

For example, in the old man's hand-shading, Father Lin on the court to stand up to face the Crow, command killed dozens of armed with a kitchen knife mop neighbor, said his classic lines, "I preach more than 20 years, so that everyone believe that Jesus will not necessarily, but let everyone chop people they will comply." ”

So take this scene and freeze it. Different media If you give this scene a title, "Reader" must be "the priest behind the philosophy of the small story", "bosom friend" must be "thick big love ah for that, Junya priest for the juvenile to prop up a day", the legal version must be "control when the tool ban?" 24 machetes were discovered in the peace neighbourhood. And on the subway street tabloid must be "The Beast Priest shows the real fangs command want to bloody Dongxing".

Data is dead, man is alive. It will be a necessary skill to learn how to interpret rationally, especially for workers like me, who must learn to find the most useful reading point of the data.

Second, from the big day two to see the first condition of the data

Big day two in the first three of the ancient puzzle is not the Mountain Dew, and the actor Xietianhua in that year also is just a junior. But at the beginning of the fourth episode, when Fei put his sister to marry him, his status within the hongxing soaring, it is this status change, let him ready to help pheasant Tunmen carry the Visigoths, failed to calmly analyze the data.

When the insider told him, Visigoths every day in the fish booth play mahjong around only three or four people, the big day two not from multiple channels to verify the accuracy of data, but heroic with four people will go to fix Visigoths, although before the departure worship Guan, but don't forget, Guan was also because of ignoring the riverside Beacon intelligence network and was cut head.

Visigoths side sure enough only three people with him play mahjong, but the room next to more than 50 people, two were pressed in the pool to pick up the hamstring, if not big fly, that night he has been dead on the street.

So, for people who study data, the right data is more useful than a lot of data. The correct here, not only refers to the correct conclusion, but also the correct sampling, the analysis of the correct direction.

In the 2013, big Data became the hottest word in the IT world, and no company could walk in the industry without saying that it would be a big data analyst. However, if blindly pursue the large amount of data, and ignore the data of the right, then it is easy to go the wrong way farther.

In a variety of data-statistics textbooks, one case has been repeatedly mentioned, that is, when Roosevelt ran for president, the famous American Journal Literature Digest released 10 million readers questionnaires, and did a lot of telephone research, and finally concluded that the Republican Langdon will be 57% to 43% Of the absolute superiority of Roosevelt over the Democratic Party. The final result was Roosevelt's re-election with a 62% approval rating, giving the literary digest a slap in the face.

In this case, is it because the number of samples is not enough? 10 million surveys were issued, 2.4 million data were recycled, and the same amount of data was today a huge one. The core problem is that there is an unconscious bias in the choice of samples, and the 10 million Literature Digest reader does not represent the characteristics of all Americans, and the 2.4 million people who are willing to participate in the survey have no similarity to the remaining 7.6 million.

From this we can see that the larger the number of sample data analysis, if there are deviations, the greater the harm caused. For example, if in 100,000 of the customers arrested by the police station to do a survey, asked whether China should open up the sex trade, then the last possible media out of the title is "Human Nature call Liberation, 99% of Chinese people strive to legalize sex transactions."

Third, from the foreskin to see the data analysis of the precise demand

In the big day two hide up injured, Dongxing Crow under the four eyes in order to find the whereabouts of two big days, sent their plump yellow hair small too sister to seduce wrapping, this trick is called precision. Because if he sent hundreds of people to inquire about the street, low cost and easy to arouse suspicion, if to seduce Chen Haonan Pheasant, they do not see this big chest and no brain of the vulgar little sister. And the foreskin? As the core of access to the information of the big day, it is the best breakthrough point.

This is before you start to do any data analysis, you must first know what you want most, get the desired results which data is necessary, do not blindly extend the dimension, which will only make the analysis results confusing.

Whether for a Web site or for an application, there will be a lot of data indicators, first want to know which of their most important indicators, and then look at the strong correlation with this indicator of which data, and finally come to the conclusion and Operation plan. For example, for the site, the boss is most concerned about the UV data, then in the plan to pay attention to the flow of the largest importer, SEO keywords, third-party platform share. If the boss is most concerned about PV data, then the operation needs to pay attention to the number of users browsing pages, user stay time, user's two jump page and so on.

I'm not a data trap that teaches you to cheat.

In many people's eyes, the data is the most impartial and the most objective and the most honest, and I wish to elevate the data to the point of worship. But the data is actually a work partner, good data is the same as God's teammates, bad data is a pig-like teammates. There is no right or wrong in the analysis of data, only the appropriate and unreasonable. (in the case of basic mathematical correctness)

Georgia, the United States, in 2011 and 2012 successively announced the 35 educators in Atlanta 65 charges, most of whom are principals, grade directors. The charges are mainly focused on cheating in exams. Because Atlanta from 2005 onwards, the student's achievement has the ladder progress, at first all all praised the newly appointed education director efficient.

But then, when the State Department of Education was doing statistics, it suddenly found that the proportion of Atlanta students ' answers was far greater than that of students elsewhere. To know from the probability, a student's answer because it suddenly through the two veins, from the wrong change is still a bit possible, but if a test room students altered traces are all the wrong change to correct, this probability is negligible.

This data is so unusual that it cracked the whole world's gaping Atlanta test fraud case. The most heinous thing is that a group of teachers held a pizza party after the exam, at the party all the students to do the wrong questions all changed to the correct.

This is the conscience of the education industry!!!

Now more and more data, traps are more and more, sometimes playing with the word game can make the performance look tall. I have a QQ group, group of a total of three men, one day this year, a tyrants friend to his wife bought a birthday gift, is a Porsche SUV, worth 1.2 million. Later, I said to my wife: "In this group, each man on average this year to his wife's birthday gift value of 400,000 yuan, what do you not satisfied?" That day, the home pans finally came in handy.

Once an application of the operation of the students wrote a weekly newspaper, "after a week of Operation promotion, the application of the download increased by 100%." I looked at the feelings of their worship, this is the tide! Later, he said to me after drinking: "In fact, I have to download the amount from the previous 25 to 50." "Suddenly, I understand, originally he just in a small QQ group to promote a bit." This shows that drinking is harmful to health.

In addition to word games, the production process behind the data is often a trap. For example, a high school science class, there are a total of 10 girls, the average cup is B, while in the sophomore year, inside a cup of girls transferred, the average Cup into C. So from the data, we can come to the conclusion that "from the B Cup to the C Cup, high school learning is conducive to the formation of girls ' body."

Similarly, the dwarf went out, the average height was raised, the poor students went out, the average score increased. So in the future when you see any data, please do not immediately cheer or dejected, think first, the data is reasonable?

I'm not a trap to teach you how to cheat.

Do a data report, not a few line chart bar chart, are embarrassed to say that they are research data. But behind every tall chart, there are a lot of little secrets to hide.

In the picture above, all the eyes are no problem people can see Tongluowan carry Chen Haonan from the "Ancient Puzzle" the third part, he participated in the number of fights and singled out a sharp decline, until 2000, he did not fight! If one has deliberately added an explanation, "the time for the fourth release of the old Puzzle is 1997". So many unreliable data analysts are about to start saying "the number of Chen Haonan shots can be seen in Hong Kong after the reunification of the public order Better" "Triad societies in the 97 after the transformation of private enterprises, financial warfare to replace the street gangs," "Hong Kong residents feel a sense of security times rise" ...

In fact, is the "ancient confused" fourth Invincible, inside charged in the frontline is and Tunmen Pheasant, Chen Haonan main time in the bubble Michelle Reis. The 2000 edition, the name is "the friendship of the days of the Pheasant story", itself Chen Haonan in here there is no drama.

Similarly, in many public relations soft article, will let the person be surprised to discover own enterprise and the competitor enterprise's trend chart scale is different, will eventually appear the enterprise's curve is so towering, and the competitor's trend is so airfield.

Vi. where to put the data

East Five Tigers in the Ben Thunder, will be in their own at the bar of the play Mozart Requiem, but this melodious music, Yao Yang will not because of his love for Mozart to give up the death of the dinosaurs.

Xiaobao will Ake as the goddess of mind, but he will not because of his worship of the goddess to give up in the big bed of the spring courtyard for Ake undress.

Also for data analysis, the data should be based on empirical understanding and scientific calculation, and then expressed and disseminated, so that it is conducive to the next decision. Many of today's data analysis focuses on computing and expression, but ignores the next decision.

Not even a vase is a data analysis that can make a decision about work decisions.

To know that data alone sometimes does not change the status quo, because people may not believe your words or data, in 19th century, Hungarian doctor Semerves through a large number of experiments concluded that "the doctors have not sterilized hands to cause the maternal death because of puerperal", but the result is one, Led to the medical authority to jump out to blame him, he thus sadly left Vienna. Until the 1865 of his death, Pasteur's germ theory came into being, and all the people faced the Semerves.

So before you do any data analysis, think about why you want to do this analysis, and how you expect to get help with your work.

Vii. the end

In fact, this article is very suitable for a name called "How to use data to deceive their boss", if you are a wage earners, then this article is only a cursory popularization of basic knowledge, more data statistics traps need you to continue to explore and improve in the work life, I believe that one day, you will be the boss nailed to the company's stigma column.

If you are the boss ... Do you see this kind of article not afraid to affect their future listing plans?

Of course, the last to give you a classic sentence of the ancient, "out of the mix, wrong to recognize, be beaten to stand firm!"