Three tricks teach you to simply predict the future.

Before dealing with all kinds of things, some people will choose an auspicious day. They used to count eight characters to see their bad luck. Now they may be very popular with constellations. Knowing their horoscope, they will help themselves with lucky time periods and lucky things, so as to predict what they will encounter and their future.

Is there a higher-order method besides that?

Yes, mathematicians predict the situation by establishing models, so as to judge the general direction of the future.

Now big data is popular, and we live in the era of big data. Through the collection of data, we can establish a mathematical model to make judgments and show us a reliable future under a large amount of data.

Although big data is closely related to our lives, as ordinary people, data collection has become a big problem, not to mention a large number of algorithms after data collection. We can't seem to use these high-end things.

Indeed, now that big data is popular, it is not practical to switch to the lives of our people. However, if there are some small data in life, we can simply process and judge these small data.

In fact, we have been using these methods, relying on intuition, but we don't know the scientific principles behind them, such as when we are waiting for the bus, judging when the next bus will come back? Is it worth the wait?

Bayesian had such doubts more than 250 years ago. How should we judge the future trend? Without the support of big data more than 250 years ago, how did Bayesian solve it?

The Bayesian problem is this:

If the probability is 100%, then the winning probability of three lottery tickets is also 100%. But if it is 50%, the winning probability of three lottery tickets becomes1/2×1/2×1/2 =1/8. If the winning probability is 1%, then the winning probability of three lottery tickets becomes111100 ×1100.

Bayes thinks that the probability of 1/8 is greater than 100%, and the probability of1100×1100 is greater than100. Bayesian also published his own paper. Bayesian's most important contribution lies in quantifying intuition and inferring excessive assumptions.

Bayesian didn't actually solve this problem, because if you ask Bayesian what is the probability? He didn't know, he just said it was more likely than that.

A few years later, the French mathematician Laplace gave a solution, and the answer was quite simple. There are w winners in any n attempts, so the future situation is: (w+ 1)/(n+2). This is the famous Laplace's law.

Laplace also used this rule in his own life. For example, what is the probability of having boys and girls in the future? Through this formula, we can get a result that the ratio of male to female is close to 1: 1.

Laplacian opened the door for our little data to deal with the real world.

Small data also has the beauty of small data. It may not be perfect, but it is perfect enough to improve our lives. After so many years of development, it can be said that there have been some phased achievements in the application of small data.

In order to expand these small data, there must be some prior data to support it. Certainly not, at least a guess, even if it is unrealistic.

These conjectures and previously verified data will determine the method we use. Three tricks teach you to simply predict the future. The following are three prior data and our prediction methods:

? Multiplication rule in power distribution

Power law refers to the development of things, and the scale is inversely proportional to the multiple. The larger the scale, the fewer times. For example, the relationship between the total box office and the number of movies, when the box office is small, the number of movies is also small. After a certain point, the number of movies seems to have not increased much, but the box office of movies has increased dramatically. It shows that some movies account for the vast majority and bring considerable box office.

Power law is also called 2/8 law. The most important part of life is determined by that 20%, and the income of investment companies is mainly brought by their investment decisions of that 20%. The total box office of movies is also mainly contributed by 20% of movies.

If you want to predict the box office that a movie may bring, it conforms to the law of multiplication. The multiplication law has a fixed coefficient, and different things have different coefficients. Suppose the box office coefficient of a movie is 1.4, then the movie has already made 6 million box offices. At present, it can be predicted that this film will have a box office of 840.

When things conform to the power distribution, the multiplication rule is effective.

The average rule in normal distribution

In the normal distribution, there are very few people at the two extremes, and most of them are in one area, so the age distribution of human beings conforms to the normal distribution. When we predict age, we need to use the average rule. We know that very young people and very long-lived people belong to a minority, and most of us will be in one area.

Suppose that the average age of human beings is 79 years old and a child is only 7 years old. How old do you predict he will live? According to the law of average, this child is at the peak of normal distribution, and we would guess that he is about 78 years old. On the contrary, if an old man is 90 years old, how old would you predict him? Similarly, according to the law, the elderly may live to be 94 years old.

A movie just came out, you haven't seen it. How long do you think you'll enjoy it? According to the average law of normal distribution, a movie is about 120 minutes. Except for a few hours like Titanic, we can estimate the time we need.

? Constant law of Ordos distribution

The law of constancy is a constant quantity and will not be changed by other influences. What we are most familiar with is that when we are playing games or immersed in our favorite things, we often say that I will be finished in five minutes, but the fact is that there are still five minutes left after five minutes, and there seems to be no tendency to stop.

In the casino, we often encounter such a situation, that is, one more hand, I don't play, I promise to go, and the result is one hand after another. Always forget what you said, which is also called the forgetting rule.

Cheng Sanfu decides Washan. Here are three ways to predict the future. Summarize these three tricks:

If you play a slot machine, and the slot machine conforms to the distribution of power, if you win once, you will probably win all the time. If you lose, don't expect to win, you will always lose. This is the law of multiplication, and this effect is constantly amplified.

If the slot machine conforms to the normal distribution, it will not always win, nor will it always lose, but in the average value of this slot machine, the number of wins and losses tends to average.

It doesn't matter how many times you play a slot machine if it fits the Erci distribution. How many times you play it, it is doomed to lose or win. Every slot machine has different settings, so it is particularly important to know what distribution a slot machine belongs to.

Steve Gould, a professor of biology at Harvard University, wanted to know how long he would live after he found out that he had cancer, and made a prediction for himself. The doctor just told him that half of the patients died within eight months after he discovered the cancer.

Gould thinks this is just one of the data, and he doesn't know the survival and distribution of this cancer.

If it fits the normal distribution, then he is about eight months. If it is in line with the distribution of power, it is completely different. The longer he persists, the longer he will live. Even with normal distribution, Gould thinks he is an extreme person and can live longer. As a result, Gould lived for another 20 years after cancer was discovered. Harvard professors also use small data to make judgments. Big data is not bad, but small data can help us when it is sometimes difficult for us to get the big data we want.

In depth, the principle behind this is actually modeling with data, and the difference lies in the amount of data.

When it is determined that things conform to a certain distribution, no matter how much data there is, the function we get is fixed. As shown in the above figure, we can get a function from several data or thousands of points, but we can get the original function with the least data.

The advantage of small data is fast decision-making, but the disadvantage is that it may not be completely accurate, which may lead to large deviation. The advantage of big data is that it is easier for us to get an accurate result, but the disadvantage is that it takes a lot of time to analyze the data.

Successful people in every industry have the ability to make decisions in a short time, and the success rate of decision-making is still very high. Why? In April this year, Harvard Business Review Network analyzed this and thought that the way to make a quick judgment on many complicated things in life is to make simple rules. Just as the three tricks teach you simple ways to predict the future, the three tricks are simplified rules so that you can judge the general direction without making mistakes.

Simple doesn't mean easy. Three simple predictions, followed by a lot of data and the efforts of ancestors. The idea of small data is so different under the trend of big data, but it is so dazzling!

reference data

Complex decision-making

https://HBR . org/20 17/05/linear-thinking-in-a-nonlinear-world dy & gt；