Speaking pit father of "big data", "data mining", "artificial intelligence" and other project requirements, must sacrifice the party looked silence, poetry Party saw tears: "Item Five afraid."
A thermal head afraid leadership
The latest technology to a
In front of the circle of friends do
What is the value of his pipe
A considerable number of "big data", "data mining" "Artificial Intelligence" project, is because the party boss of a department to cater to the meaning, the latest fashion, it was on the line . Anyway, the big boss likes, though he have any business scenario, though he have any commercial value to dry up.
This project if lucky, to catch up with rising performance of the company, may be responsible for a meteoric rise, to leave his ass next rub.
If bad luck, catch the falling performance of the company, it will not even responsible for the project together with swept into the dustbin.
This is known as the mascot project . As a party, if followed by a hot head when the mascot, and most likely are not being paid the balance due. Because in essence, the data is difficult to directly generate the value of projects, the need to combine a business scenarios. So before the project is necessary for the project to find a way out of homework.
(Over the air network red to waste, no practical application scenarios, the application is not enough just to be, the scene is not important, the final fate of the project are better where to go)
Second, fear not carry clear business
Data expect cure all diseases
He needs to write two sentences
Grabbed problems
A funny argument is: a problem in the eyes of the business sector excel pivot tables can not be considered a great deal of data. Combined with years of marketing No. fierce advocate, caused a lot of business on the "big data", "data mining", "artificial intelligence" had unrealistic fantasies. If demand is business to mention, in particular, you have to think: they mention in the end is what needs? This requirement has no conditions to do? To the extent of how much data can help?
How to sort out the demand, expand are enough to write a book, just to prompt some frequently asked questions:
He or she says , oral questions and the actual data are two different things.
Ambitious , far exceeding the expectations of the actual IT capabilities, not willing to put money.
We expect too much to expect miracles data. To a "big data" a practice, not a backlog of inventory, customers answer the phone, there is no risk, the results have friends.
These questions come from the business sector is too brightest marketing number brainwashing, but it do not understand technology. When talking about the needs of the business written on paper a punctuation mark had to be sure. Control expectations from the outset, to find a reasonable way to the floor, is the jumping pit positive solution.
Three fears lazy line execution
Data Entry nobody
How how to do it the easy way
Full data quality finished
A lot of owners are referred to the data cleansing, interesting is: quite a lot of dirty data is artificially get out. The most typical, such as shopping in a supermarket, the cashier questions: "? You have a membership card," you say: "No", he would not care ...... This may seem a little negligence, resulting in a lot of business over the data associated with the customer ID less than 10%, the data base is so bad, in addition to what the whole order analysis can analyze? It lost 90% of the data model and build out response dim?
Not to mention what channels to manipulate results, counterfeit cards, cards ghost, ghost card, wool off and so on. I encountered a bank clerk taught me: "! Here you do not fill more than 10w annual salary nor less" thing. In fact, data management companies do well, in itself force control and execution are very strong. If a company's control power shortage, the data items that do colleagues, promised at the time of data quality project, we have a mouth mercy.
Four did not support afraid floor
Expect to eat data
Business Unit inaction
Where's performance jumped up
Ultimately, the data item you want to promote business growth, the business sector is the need to cooperate. If a marketing class project, required products, advertising, promotional costs; if production projects, require manual operation, workflow, job training, and implementation of norms. This is a very simple truth: even the best sniper shooter have to pull the trigger to kill the enemy.
However, many business, not too superstitious data, the data is too lightly. Scientific data, such as: "Why are precise push the products have good targeted marketing of why investment promotion??" Contempt of data, such as: "the headquarters of bullshit, our own doing!" So it goes, plenty of data, there is no ground, how to reflect the value of the data?
Now, however, a large number of newcomers also contributed to this trend. Do not consider the business model of the scene, looking for a cleaned packet data directly lost inappropriate, then the blind pursuit of the test set to run out accuracy. It does not seem to understand the business packets can have 95%, so there will be a fifth afraid:
Five afraid of the big newcomers mouth:
"I know I am not afraid algorithm
He pipe problem is Gesha
Code for a run to get it! "
Is a drop, veterans do know that deep water projects. If demand is the IT department to mention, it is also relatively tricky, there are at least skilled people to help us filter once. But if it is business, it really has the spirit of playing 12 minutes. However, geese, new people are often a hot head on the last, what are like nothing complicated. The results do, or abused convinced, either abused departure. As the saying goes: New Ghost tired old injustice Guiku. Students joined the line is not abused several times, it is difficult to appreciate the one taste of.
(Often lifted an example, even though, can be a preliminary estimate of the prospects of the so-called precision marketing through simple descriptive statistics. Do not underestimate Oh descriptive statistics, histograms in addition to be considered a frequency, which is fast approaching business, an important means of pit excavation)
Tucao Tucao go, after all, or would like to remind you the importance of doing business close to the project. Because in essence, data services Yu Renshi work of a professional trades. Just as a doctor with a patient, doctors are better educated than Dr., be patient but it is part of everyday kinds of colors. Most of the patients, the door is only hum: Yo eh, eh Yo, uncomfortable, uncomfortable. This time you have to be patient to diagnose the illness, not that I have a alpha hormone, a needle is not going to cure the patient.
Author: Chen grounded gas
link: https: //www.zhihu.com/question/266405986/answer/342445055
Source: know almost