How does data mining actually work?

Suppose I want to do some data mining on the database of a supermarket. What does that actually mean?

1) What will the output/results be like?

2) Will the output be different every day or change over time?

3) Before applying data mining, do I need to know what I want or will data mining give everything I w开发者_Python百科ant automatically?

---

Data Mining is a general category of techniques that can be applied to different kinds of datasets, just like programming is a general category of techniques that can be applied using different languages to do different things.

None of your questions make any sense.

---

A1: Data mining will give us an accurate reports about your queries of database of supermarket.

A2: Sure, because Data mining depend on analyzing during time, in this case it depend on your problems or goals that you want to reach it. if your database was very big also you built data warehouse in right way you will get the different output over time.

A3: yes you should determine what are the problems you have to mine then use tools of Data mining to get the results or indicators automatically.

---

To answer your first question: For the case of supermarket customer data, I could image the following questions:

• how many products X are usually sold on Fridays ?

  (helps you to determine how many X you should have in stock)

• which customers bought product X often in the last month/year ?

  Useful when when you introduce a new X-like product: send advertising material (which has a given cost) only to those customers.

• given a customer buys product X (e.g. beer) what's the probability that he/she also buys product Y (e.g. chips) ?

  useful for the following: make sure X and Y never are on promotional offer at the same time (X and Y are bought together often). Get the customers into the store by offering a rebate on X knowing they'll also by Y at the same time. Or: put a high price X-like product right next to Y, putting the cheaper X somewhere else.

• which neighborhoods have the smallest number of customers ?

  helps to find out which neighborhoods you could target with advertising to bring more customers into the store.

Often, by 'asking certain questions to the data' one discovers some features and comes up with new questions.

---

Data mining is a set of techniques. It refers to discovering interesting and unexpected patterns in data.

If you want to apply some data mining techniques, you need to know which one and you should know why. The answer to questions 1, 2 and 3 depends on the techniques that you choose.

For example, if i want to find associations between items sold in a supermarket, i may use association rule mining. If i want to find groups of similar customers, I might use a clustering algorithm. etc.

There is not just ONE technique in data mining.