This section focuses on applications of information theory. This section introduces big data and Internet advertising.

1. Grasp the essence of big data from the four characteristics

MeiTuXiu XiuXiu repair and use Huawei cell phone to take pictures are the result of using big data.

How can you develop a pair of eyes to determine whether a company is really using big data? That starts with the characteristics of big data.

First of all, big data requires a large amount of data, and there is no question about that. A small amount of data must not meet the principles of big data. As for how much data is appropriate, we introduced the concept of confidence earlier, the data should be at least large enough to allow the results of the statistics to have a very high degree of confidence.

Second, big data needs to be characterized by multiple dimensions, and the dimensions should ideally be orthogonal.

The third characteristic is data completeness. It was often overlooked in the past because humans used data in the past, using a sampling approach to obtain it, and it was simply not possible to achieve completeness. Today, the situation is different, because the devices that collect data are ubiquitous, and we are consciously or unconsciously feeding data to it, so it is entirely possible to obtain complete information, thus blocking the dead end of the use of data for prediction.

The fourth characteristic is that big data also needs to be real-time, because in those application scenarios where a certain amount of time passes, the data loses its meaning.

2. In the coming era, there are three types of businesses that will benefit from big data. The power of big data we all see, then there may be some people want to ask, with big data, I was engaged in the industry will disappear? In fact, most of the industries will not disappear soon, but may appear in another form. Eliminating contradictions between data also requires domain knowledge. Therefore people who have been in an industry for a long time and have specialized knowledge will not only not be replaced by big data, but also have the potential to make good use of big data and move up the career ladder. There are three types of businesses that will benefit from big data in the coming era.

The first type of enterprises are companies similar to Tencent, Alibaba, or Today's Headlines, which have their own data, technology, and application scenarios, and don't need help from outsiders.

The second type of enterprise has data but no technology, including many large enterprises, such as mobile operators, traditional banks and retailers, etc. They need help from outsiders, but usually do not cooperate with the first type of company to protect their own interests. The professionals in this category will be able to solve the problem of data conflicts above, as well as linking different dimensions of data within the industry.

The third type of company lacks data, but has the technology, so they need to solve real problems for the second type of company. Of course, in the process of solving the problem, they may be able to enter many fields, and gradually become a new platform company, such as the famous U.S. big data company Splunk is so step-by-step development, it is today's market capitalization of up to $ 20 billion. But in the beginning, this type of company must be doing the dirty work.

3, the four levels of big data thinking

We used to say, quantitative change will bring qualitative change, that is often said in one dimension, and today we say that big data thinking, has exceeded this layer of meaning, is a new way of thinking and doing things.

The first meaning of big data thinking: from a large number of seemingly haphazard data points, summarize the original correlation can not be found.

The second layer of meaning of big data thinking: do not make assumptions in advance, from the big data to get the conclusion first, and then analyze the reasons.

The third meaning of big data thinking: the use of big data to accurately control the macro law at the same time, accurate to every detail.

The fourth layer of meaning of big data thinking: is through the strong correlation of several dimensions, instead of the past cause and effect relationship.

The four levels of big data thinking, they are progressive, I hope you keep in mind, think about how you have big data thinking.

1, Google's advertising system utilizes user-initiated information

User-initiated information is the most effective, because at any time, people always like to buy things, do not like to be sold things. No push is more effective than a user-initiated request.

2, the difference in people's preferences for various things is much smaller than we usually think

The essence of personalized service is to find the degree of difference between each person and the whole group in their preferences, and we can use cross-entropy to quantitatively measure it. The finer the dimensionality, the more prominent the personalization characteristics. However, the differences in people's preferences for various things are much smaller than we usually think.

Today there are many entrepreneurs who don't pay attention to the underlying technology and information theory, and naively think that if they do something personalized, they can gain a crushing advantage over existing industry competitors, which is just their own imagination.

Google has always emphasized that you can't draw any conclusions without data, and the reasoning is similar. In the information age, why do we need to understand the most basic principles of information theory? It is to be able to judge whether the direction of doing things is correct, so as not to die like some companies do not know how to die.

3, the use of orthogonal, stackable information has a great role

This is the fundamental reason for the success of Facebook.

4. The shorter the link to closing the deal, the more effective the ad is.

Most media, including information on portals, just provide general information, they are the furthest away from the purchase, so the ads are the least effective. The information on social networks and Google search belongs to the second and third stage, which is closer and closer to the purchase, and the effect of the advertisement is better and better, and the one on e-commerce belongs to the last link, which is the most effective. I have said on many occasions, do things straight, the best results, do not beat around the bush, this is the reason.