A few days ago, I wrote a short article entitled “Please do not confuse Machine learning with Data Analysis”. Many partners who are interested in machine learning sent me private letters asking me: They talk about machine learning all the time, so where is machine learning applied? In fact, the application of machine learning is too extensive, so I hope to share a few typical applications of machine learning with you through this small article today. Look at how machine learning is permeating every aspect of our lives, affecting the way we eat, dress, and live.
One of the earliest and most famous cases is the case of “beer + nappies”, for this case, anyone who is engaged in data related work should know this case, because it is so famous. Whether you’re a machine learning person, whether you’re a data mining person, whether you’re a data analytics person, everybody’s talking about this case. So without further ado, let’s see what this case tells us.
At Wal-Mart, America’s biggest supermarket, data analysts often find that beer and nappies are two very different items that customers often buy. In other words, after a person buys beer, that person is more likely to buy diapers. After this person buys the diapers, he’s likely to go out and buy beer. This phenomenon caught the attention of the walmart staff, who then went out to investigate. After the survey, we found an interesting phenomenon. We should all know that the large supermarkets in the United States or Europe are not located in the city center or densely populated areas like some of our domestic supermarkets. In foreign countries, many large supermarkets are located in the countryside or some remote suburbs far from the city. For many ordinary families, most of them make a big purchase every time they drive a car. Then the data staff found that for many families with newborn babies, a lot of purchases are made by the fathers. Then when these fathers buy diapers for the new members, they will also give themselves a reward and buy some beer, which is the reason for such a phenomenon. So after the walmart staff found this phenomenon, they made some interesting sales strategies in sales, such as adjusting the position of the shelves and making a bundle sale of the two items. The net effect was to increase sales of both products. That’s the case with beer and diapers. So this is our first example of turning data into money. And this case, from our algorithmic point of view or our application point of view, is called shopping basket analysis.
So what’s shopping basket analysis?
Basket analysis is a machine that looks at which items in our order are purchased at the same time, and the algorithm used here is association rules. So in a sense, this association rule or shopping basket analysis algorithm is not a typical machine learning algorithm but a typical data mining algorithm. But for the people of our application, we don’t care which school it belongs to, as long as you can help me turn the data into money
So, this shopping basket analysis is also an algorithm that our current machine learning buddies should know.
So let’s look at the second example
The second example is targeted marketing
This user segmentation precision marketing specific significance we will not say, I now cite a specific case. So we know that in previous years, the mobile numbers we used were actually different brands. But now the movement doesn’t seem to be so divided. So in the past, he divided so many brands, each brand has a clear positioning. Such as global, global is the main concern is the so-called high-end business people, these people may be flying all day at home and abroad, and these people for your calls cost a few cents a minute is actually don’t care, he is concerned with this brand is for his own what additional value-added services. For example, we often see global VIP lounges in airport lounges. This is the value-added service brought by this brand. At least, these users feel they have face.
Then I remember and m-zone, m-zone mainly aimed at as if those students, at that time we all know that time is not smart, not to mention what WeChat IM tools such as weibo, so that when students in the main way to communicate is the text, they call for the function may not be very strong, But the need for text messaging can be hungry. They may send a lot of text messages, and many of them are of little substance. Just asking where you are? Can you open the door for me? Have you eaten yet? All of these have no content, so this dynamic zone is aimed at this kind of school students to design, he this package may include a lot of traffic package ah, SMS package ah.
Another brand is Shenzhouhang
Shenzhouxing is mainly aimed at ordinary people like us, because we may have more needs for phone calls, such as frequent practice with customers, often call home, some long-distance needs, he will have a big discount on the cost of the call.
This is a typical case of user segmentation. But the question is how does he come up with this idea of user segmentation? Here, from the perspective of machine learning, clustering algorithm can solve this problem completely. Clustering is a typical algorithm for this kind of problem. For example, we take a large number of users’ data and feed it to the clustering algorithm. Without our deliberate manual intervention, we completely rely on the computer to do this calculation. Finally, the computer will divide the data into several categories for you, and you can divide them into several categories if you want. After good classification we can let business personnel to analyze which kind of users have what common characteristics, and then extract these characteristics, let these business personnel to sum up what characteristics these people have, and then make a few packages to cheat you. This is a typical case of user segmentation
Ok, I believe that through this case you can understand our machine learning will have so many daily applications, also welcome you to pay more attention to.
Besides, thanks to everyone who tipped and liked last time, I can buy several packets of latiao to calm down the shock. Thank you very much!