One case study to understand the entire process of data analysis

Through data analysis, people can identify trends, discover relationships, and predict future trends, helping to make smarter decisions and develop more effective strategies. This article describes the entire process of data analysis, and I hope it will be helpful to you.

New students often ask: What is the complete process of data analysis? Today, I will use a simple example to help you understand the complete process of data analysis. By comparing, you can also find out why you feel that you have not done a complete data analysis.

One day, your friend Lao Wang said to you: "Why do you still work? You only make 10,000 yuan a month at work. You might as well sell pancakes like me. I can earn much more than you do at work!" You were surprised and curious. Does selling pancakes really make much more than working? In order to answer this question, you have to do an analysis. What Lao Wang said to you before, in professional terms, is called: understand the analysis background .

So, how do you determine whether selling pancakes or working as a worker earns more? You may intuitively think: it is clear how much money you earn as a worker in a month, but it is unclear how much money you earn from selling pancakes in a month. So, you write down a comparison table as shown below, and then start collecting data. This action is called " determining the analysis target " in professional terms.

Although the hand-pulled pancake stall is small, it sells a lot of miscellaneous items.

• Income: Selling cakes, sausages, chicken strips, drinks...
• Cost: cart, stove, gas tank, sauce, chopped green onion...

It is too tedious and difficult to count them one by one.

Therefore, you decide to simplify things and focus on the most important costs and revenues. You go to Lao Wang's stall and learn the following key information:

1. The most basic original flavor hand-pulled pancake: 1 pancake + 1 egg, 5 yuan

2. Sauce and chopped green onions are small expenses. The main costs are pancakes, eggs, and natural gas.

3. The ham sausage is purchased at 0.2 yuan and sold at 2 yuan, and the chicken fillet is purchased at 0.3 yuan and sold at 3 yuan, which can contribute a lot of profit.

4. The other bits and pieces are just scraps, but better than nothing.

So, you organize the following logic diagram and data table. This step, in professional terms, is called building an analysis model.

After doing this, you can start collecting data.

1. WeChat and Alipay are used for payment, so you can directly view the amount of money received on the day.

2. The consumed biscuits, eggs, ham sausages and chicken fillets can be counted once every morning and evening.

3. The gas consumption cannot be calculated on a daily basis. You can only calculate how many days it takes to replace the gas twice and allocate it.

This process, in professional terms, is called data collection . During the collection process, we should pay attention to eliminating some problems, such as losing 1,000 yuan playing mahjong at night, which has nothing to do with the stall, so it is excluded from the WeChat income and expenditure. This is called data cleaning .

After collecting detailed data, you can calculate the daily income and expenditure, as shown in the following table. This process is called data calculation (commonly known as data acquisition) . The complexity of data acquisition is directly related to the complexity of data calculation. If the data is as simple as the above, the work will definitely be much easier.

There is a poem that goes: "A cup of tea, a cigarette, and a piece of SQL to write for a day", which is complaining that data calculation is too complicated, resulting in hundreds of lines of SQL to be written.

After seeing the data, you will be curious: Why is Mr. Wang's income so unstable? One day it is high and the next day it is low. After understanding the details of each day, you find that:

1. On rainy days, there are fewer people on the streets and income decreases (external factors)

2. I got up late today, failed to get a good seat, and my income dropped (internal factors)

3. I had a fever today and felt uncomfortable, so I only worked until 9pm and closed the stall early, resulting in a decrease in income (internal factors)

It seems that if you don't record these reasons, you can't do in-depth analysis. So you record Lao Wang's internal and external factors every day as shown in the figure below. This action is called adding analysis dimensions (commonly known as labeling).

With the analysis dimension, we can explain why Mr. Wang’s income is unstable, and we have a certain ability to predict. For example, if you see the weather forecast and it is expected to rain for two weeks this month, then Mr. Wang’s income will definitely not be maintained.

Combining all the above information, after 1 month, you finally completed the analysis goals listed at the beginning, as shown in the figure below.

Overall, it seems that Lao Wang's salary is indeed higher than yours during the statistical month. However, after careful analysis, you understand the factors that affect Lao Wang's income fluctuations, which may add new dimensions to your judgment. For example:

1. Can I persist in setting up a stall 28 days a month?

2. Can I get up at 6am to grab a good seat?

3. Can I hold out until 9pm even if I have a fever?

This process, in professional terms, is called: adding evaluation criteria .

Finally, you come to the conclusion that although the income from setting up a stall is high, it is too physically demanding and has poor stability, so you refuse to accept the suggestion of setting up a stall. This is called: drawing an analytical conclusion .

The above is the whole process of data analysis:

1. Understand the analysis background

2. Clarify the analysis objectives

3. Build an analysis model

4. Data collection, cleaning, and calculation

5. Add analysis dimensions and evaluation criteria

6. Draw conclusions and suggestions

The common reasons why many students feel that they have not done a complete analysis are:

• As a grassroots employee, I only do step 4. I don't know why the numbers are collected and what the numbers are used for.
• Without a clear goal (step 2), no conclusion can be drawn after collecting data, and no effective suggestions can be made.
• I don’t understand the business, I don’t know what the model is (step 3) and I don’t know which dimensions are useful (step 5)
• The company's data construction is too poor, lack of data, poor data quality, and consumes a lot of energy (Step 4)
• The business department doesn’t cooperate, and my own leader is too lazy to teach me, just telling me to “think more”…

Author: Down-to-earth Teacher Chen, Source: WeChat public account "Down-to-earth Teacher Chen"