How to cultivate data thinking

Whether there is any industry in the industry, data analysis is increasingly becoming an essential skill, while using data thinking to make decisions can produce high quality decision results. This article combines yourself with your experience to share data analysis methodology and the principles of two data thinking, and hope to help you.

Hello, everyone, I am a little!

In recent years, a concrete data analysis work has not been engaged in the work, but it is impossible to deal with the data to be dealt with, and good data analysis thinking is possible to form decision judgment.

This article briefly shares the methodology of some data analysis, combined with deliberate practice in daily work, gradually forms more than 90% of data thinking capabilities.

In the face of problems, the usual ideas are zero. Methodology is the method of “canify scattered ideas into organized analytical ideas”.

Simple summary of the data analysis methodology of the data, it is known to know the weight, there is a strike, dismantling, and more comparison.

First, know

If a person can do it in the world, it can be said to be a mature person. Like data analysts, everyone’s energy is limited. If you choose an unimportive topic to study , Consume a lot of energy, and the results may have a little effect.

It is recommended to divide the problem by matrix analysis. The method is divided by the property A as the horizontal axis, the attribute B is the longitudinal axis, which constitutes a coordinate system, and the scale division is divided on the two-axis shaft, constitutes four quadrants, will Each of the analytics should be projected to these four quadrants, and cross-classification analysis, intuitively manifesting the association of the two attributes, and analyzing the performance of each thing in these two attributes. The classic is the Boston matrix model.

The essence of matrix analysis is a combination of different dimensions, which can be two-dimensional or three-dimensional, four-dimensional or even more, depending on the complexity of your analysis.

Through matrix analysis, we can face our problems, rush to sort, and to deploy the corresponding strategies, which area is placed in which area, which determines the ceiling you achieve.

If the famous name of the chain home, it is difficult to do it.

Back to the data analysis, if you see the indicator volatility, do you want to analyze it deeply, at this time, you can fluctuate as the horizontal axis, whether it trends as the vertical axis.

There is no doubt that the upper right is the most important and most urgent need for solving, but in daily work, it will generally do not appear frequently, and the upper left corner is a need for continuous investment.

Second, there is a strike

By knowing the topic of the analysis, it is possible to divide the topic of the analysis, determine the priority, the second step is to start analysis, and the problem that is often encountered during the data analysis process is that there are too many data, which leads to the process of the analysis. The goal, drilling into the horn tip, spending a lot of energy but has little effect.

What is more ideal is to use similar thinking map tools, surrounding the analysis objects, the direction of the analysis is exhausted, summarizing the analysis process, surrounding the thinking map, avoiding where to analyze where to analyze the Buddha analysis, and avoid The thin branches are poor, and there is a waste of resources.

When you draw a thinking map, you recommend MECE analysis, full name Mutually Exclusive Collectively Exhaustive, meaning “independence, completely exhaustive”. That is, for a major topic, you can do non-overlapping, unaffected classifications, and can take this effectively to grasp the core of the problem, and become a way to effectively solve problems.

For example, a diba is analyzed how to increase loan balance, loan balance = loan delivery – repayment – verification, then surrounding this basic formula, ensuring that the analytical dimension is not heavy.

Third, can disassemble

Everyone’s dismantling ability determines whether he can effectively handle and solve complex matters, the disassembly ability is simple, is to disassemble a complex issue into a foundation element, you use these elements, control and change basic elements. In turn solve complex problems.

The following introduces two basic dismantling methods:

Dimensional subdivision

Separation is briefly, that is, split according to each of the different dimensions, and the largest variational field of change in change. For example, if the guest declines, it can distinguish between the decline in online channels or the offline channels, and further discovery is that the specific channel declines, which is more targeted.

Separate thinking, in addition to the positioning problem, more importantly, if we only look at the overall data, it may notice the difference between the components of the data within the data. “

For example, this month is 50,000 / flat, last month’s new room 20,000 / flat, everyone saw exclaimed housing prices, but if the structure is split, it will find 80% of this month’s transaction new house. 80% concentrated in the central ring, and last month 60% is except that the outer ring is completely unchanged, and the impact of the trading structure has been misleading.

2. Process subdivision

The most intuitive is a funnel chart. The funnel map is a comparative specification suitable for business processes. The cycle is relatively long. The various process links involve more management analysis tools in complex business processes. The funnel map is the most intuitive expression of business processes, and Also the best explanation of the problem.

For example, guests are a long-term business scene, involving electric sales staff, customer access, customer wishes, filling in information convenience, approval efficiency, approval rate, user activation, etc., through funnel The figure can quickly discover the links in the business process to determine the business bottleneck. Four, more comparison

Finally, it is more comparison, just as we say, there is no harm to the comparison, compare a very important part of data analysis, and introduce several common comparison methods:

1. Compared with the same industry

Generally, it is to clarify the gap between herself and industry benchmarks.

In general, the data source of the same industry mainly discloses the data, including the financial report of the listed company, the active disclosure of data, etc., the data from the information source will also have various differences, but by careful analysis or some What you want, depending on the ability of each person’s data thinking.

2. Compared with yourself

In general, it is a common comparative idea to observe the trend of the year, the trend graph, but pay attention to the initial set goals.

It will be found in the work. Sometimes we will find that the indicators have risen over the same year, and the indicators have risen, create a thrilling situation, but in fact, there is no achievement of our goals, but the reference value is too low, this is a typical Goal erosion.

3. Subtrance comparison

Simply, in accordance with various dimensional subdivision, for example, according to a feature, divide the data into different groups, then compare the data of each group, such as a variety of similar groups often change at work. The product will continue to be iterated with your development and testing, which leads to a different experience in the first week of the product released and the users who have been added later.

Each group of users constitutes a single group, participate in the entire experiment process, by comparing the different group groups, you can learn about whether the performance of key indicators is getting better and better.

The above is some simple data ideas sharing, and finally share the principles of two data thinking:

(1) For data industry practitioners

To be sensitive, it is necessary to be able to be reasonable, the data that is not blindly believes, but the data is accurate, especially for some abnormalities, which requires time to cultivate data common sense, learn, can only deliberately practise.

(2) For some external data

It is very difficult to get it, and it is not too presented. No one has the perspective of God, calm down the data you can get, may have a deviation, but the trend, principles and opportunities behind the data, these defocated people.