As we all know, mastering Excel does not mean mastering data analysis. Just because you can tell the case of beer and diapers does not mean that you can understand the data, and PPT can not add points to your data analysis ability... We do data analysis in order to analyze business problems in a quantitative way and draw conclusions. There are two key words: quantification and business.
How to become a data analyst?
Quantification is to unify cognition and ensure that the path is traceable and replicable. After unified cognition, we can ensure that people at different levels and departments have equal voice and discuss and collaborate in the same direction, so as to avoid people in the company judging the current business situation with "I feel" and "I guess". Ultimately, only by solving business problems can we truly create value.
Many colleagues and friends have been asking me how to become a data analyst and data product manager? What knowledge reserves are needed? Today Xiao Yi comes to systematically share what is a data analyst and how to learn these skills and knowledge quickly and deeply?
—01—
What is a data analyst?
Data analysts refer to professionals specialized in industry data collection, collation, analysis, and industry research, evaluation and prediction based on data in different industries.
In other words, data analyst is a post. As long as he/she has mastered the data analysis methods and thinking, and is engaged in technology and business, he/she can be called a data analyst. In essence, his/her work content is to analyze business value or model and discover knowledge from data, so as to promote business and assist decision-making.
—02—
Data analyst
How to create value for the enterprise?
A complete enterprise data analysis system involves multiple links: collection, cleaning, transformation, storage, visualization, analysis and decision-making, etc. Among them, the work content of different links is different, and the time consumed and value generated are also far different, as shown in the figure.
For example, there are at least three aspects of data in the Internet enterprise data analysis system: user behavior data, transaction order data and CRM data. The engineer collects data from different sources, and then unifies them to the data platform through cleaning, transformation and other links; The data is then proposed from the data platform by a special data engineer. These works take up 90% of the time of the whole link, but only 10% of the value is generated.
The data analysis of this pyramid is closely combined with the actual business, and supports the business decision-making of the enterprise in the form of reports, visualization, etc., covering all front-line departments such as products, operations, markets, sales, and customer support. This part takes up only 10% of the time of the whole link, but it can produce 90% of the value.
An excellent business data analyst should be value oriented and closely combine product, operation, sales, customer support and other practices to support each business line to find and solve problems and create more value.
—03—
Common categories of data analysts
What are they?
1. Data Product Manager
Increase data thinking based on the ability of product managers. The data product manager not only understands the principle of buried point, but also can capture data and analyze it through tools such as packet capture. At the same time, they can also participate in the production of data products, such as BI reports, CRM systems, AB test background, etc.
For example, if the boss wants to build a user behavior monitoring platform, it needs to be converted into detailed technical requirements according to the boss's needs, and then submitted to the technology for development. This is the daily work of the data product manager.
2. Data analyst
This is what we often call commercial data analysts. They are mainly responsible for building 0-1 visual monitoring reports, using data mining and business insight to provide data support, analysis reports, commercial models and other services for demand departments. The core of this is monitoring, mining, value and services, and play the role of the eyes and brains of leaders in the company.
3. Data Modeler
The data modeler, also called algorithm engineer, is a model master integrating mathematical statistics knowledge, programming and business thinking. By establishing mathematical models and using algorithms to achieve growth, he can be said to be the soul worker of a product, such as the recommendation algorithm of information flow product, anti fraud and credit rating of the financial industry.
4. Data Engineer
The function of the data engineer is more inclined to technical engineering. The main responsibilities are to build a data warehouse, create an ETL, conduct data governance, data security and other work, improve the running speed, optimize the data structure, and better serve the data users, such as data analysts, data product managers, and data modelers.
5. Data scientist
Comprehensive talents, talents with data analysis ability, statistics foundation, business ability, algorithm and communication ability. Including all the above technologies and capabilities.
—04—
Become a data analyst
What capabilities are required?
1. Business capability
In the end, only by solving business problems can data analysis truly create value, that is, data analysts need to have business capabilities, and each business of an enterprise is essentially the support of the company's overall strategy, because data analysts must first understand the strategy to choose the right direction for analysis.
Secondly, you should be sensitive enough to your own industry and fully understand the industry. That is, more communication with the core team of the business department, more attention to the industry website, more reading of the industry data analysis report to accumulate, such as what stage you are at, where you are, where the current key business direction is, what challenges you have encountered, and what the overall solution is.
Finally, you also need to have practical experience in business positions. The understanding of business is not simply based on documents. It must come from the full understanding of the actual processes, mechanisms, platforms, data, etc. of the company's business. It is best to practice in the actual relevant positions.
2. Data capability
As a data analyst, you need to first understand the enterprise's data indicators. Each enterprise has a set of KPI indicator system, and there are a series of implementation monitoring indicators around the KPI indicators. As a data analyst, you must have a deep understanding of the enterprise's core indicator system. To be able to distinguish the differences of indicators in essence, you must have a thorough understanding of the indicator generation process, including which table to use, Which field is calculated and summarized layer by layer.
Secondly, you should have a global data perspective, that is, in most companies, the work of data analysts is professional, but in fact, the data you want to analyze is comprehensive, and there is no defined professional boundary. In practice, data analysts often do not know how much data they have. The depth and breadth of their data analysis are limited by their narrow vision. Data analysts should systematically learn the data dictionary. Bottom up practice is important, but top-down learning is also necessary.
Finally, you need to have a deep understanding of the data. The data dictionary often reflects the meaning of the data on the surface. If you want to analyze more flexibly, you need to understand the dependency and context between the data, because each data table is related to the next level of tables, but summary means the loss of information, and you only have the ability to trace the source, You are more likely to get more freedom of analysis based on more information, such as the function of a menu on the business system, and what data needs to be mapped to the system.
3. Technical capability
As a data analyst, you still need to have the necessary skills, such as proficient in SQL, database principles, Excel/report/BI tools. In addition, upstream and downstream technology fields, such as data warehouse, data architecture, and ETL, need to understand and even use, such as:
(1) SQL is the most flexible language for operating data. Any database will provide SQL support. It builds a bridge between business and data. It is easy to learn, cost-effective, and must be learned by data analysts.
(2) EXCEL provides the most flexible ability to process and present lightweight data. Mastering EXCEL is the basic skill of any data analyst. Perspective, chart, formula and calculation are extremely convenient tools.
(3) BI is, to a large extent, the art of using some visual techniques to compare indicators, which helps you find and locate problems more quickly and intuitively. After all, the human brain is more sensitive to charts and images.
(4) Data mining technologies, such as clustering, classification, prediction, etc. With the reduction of machine learning and artificial intelligence tools, data analysts should master at least one mining method. Know how to build models, especially in industries with high data maturity, such as finance, operators, the Internet, and retail.
4. Communication ability
For data analysts, communication ability is very important, because many projects need to be promoted by the upper level, and then the leaders of all business departments need to cooperate with you to sort out the data in the demand, while the implementation requires the cooperation of the whole chain of technology and business.
The essence of communication is to solve problems. It is not so difficult to understand the purpose of communication, express logically and clearly, and then stand on the other side to consider what the other side wants.
For example, for upward communication, we should seize every opportunity to communicate clearly what the purpose of analysis is and what the leaders expect. At the same time, you also need to face different positions, meet different roles, use different languages, express your requirements and obtain what you need, such as how to understand the business? How to get data faster? How to identify the causes of data problems as soon as possible? It tests your actual contacts and authority.
In addition, data analysts also have an important expression, which is to report data analysis results. They should learn to connect problems with analysis scenarios and tell stories. They should be able to promote the value of data through quantitative figures and vivid scenarios.
—05—
How to quickly become a data analyst?
1. Excel data analysis
Every data analyst is inseparable from Excel. It is the most commonly used tool in daily work. If performance and data volume are not considered, it can handle most of the analysis work. Although machine learning is common today, Excel is still the undisputed first tool. Excel is a tool that must be skilled for you who have no experience. It is the most commonly used tool in daily work. If performance and data volume are not considered, it can handle most of the analysis work.
2. SQL Database Language
As a data analyst, we first need to know how to get data. The most common way is to get data from a relational database. Therefore, you can't learn R or Python, but you can't learn SQL.
In the DT era, data is growing exponentially. Excel has no problem in processing data up to 100000 entries, but to a small extent, if the product has a small scale, the data will be millions. At this time, you need to learn about databases. For example, in the recruitment conditions of many enterprises, more and more product and operation posts will give priority to SQL as a bonus item. SQL is one of the core skills of data analysis. From Excel to SQL is a great progress in data processing efficiency.
Mainly understand the database query language, where, group by, orderby, having, like, count, sum, min, max, distinct, if, join, left join, limit, and or logic, time conversion function, etc. The fastest way to learn SQL is to download database management tools and find some data to practice. The client recommends MYSQL here. Recommended book: MYSQL Must Know and Must Know
3. Data visualization&business intelligence
Data visualization is not only a technology, but also an art. The same data in the hands of different people will show different effects. Mastering this technology will become a bonus in the workplace.
4. Mathematical Statistics
Statistics is one of the most important foundations of data analysis and the cornerstone and methodology of data analysis. Statistical knowledge will require us to look at data from another perspective. When you know how stupid it is to see the difference between the two groups of AB by the average value, your analysis skills will also be significantly improved.
Here we need to start from the basic statistical theory (descriptive statistics, interval estimation, hypothesis testing, etc.), to the basic statistical analysis (T test, ANOVA, etc.), and finally to the commonly used commercial models (regression analysis, ANOVA, etc.), learn the logic behind data analysis, master the concept of practical statistics, and be able to use statistical thinking to think about problems. Recommended books: "Statistical Basis of Data Analysis from Zero" - Cao Zhengfeng, "Statistics" - Jia Junping
5. Data analysis and software application
SPSS is an introductory software for statistical analysis. If you want to get started quickly but do not want to learn programming, I recommend using SPSS. SPSS software is one of the three largest statistical analysis software in the world. With its advantages of easy operation, easy entry, and easy reading of the results, it has always been favored by data analysts. Generally, after short-term learning, SPSS can be used to do simple data analysis, including drawing charts, simple regression, correlation analysis, and so on.
The focus of learning SPSS is not the software itself, but the relevant statistical knowledge, which is also suggested to pave the way ahead, that is, you should learn how to analyze the "results presented to you by the software after data input". Recommended books: SPSS/SAS EG Implementation of Better Data Processing - Xu Xiaogang, SPSS/SAS EG Advanced for Data Analysis - Chang Guozhen, Basic Course+Advanced Course of SPSS Statistical Analysis - Zhang Wentong
6. Data analysis industry application and data analysis thinking
For data analysts, business understanding is more important than data methodology. Of course, it is a pity that there is no shortcut to business learning. Recommended books: Growth Hacker, Lean Data Analysis
—06—
Summary
The daily work scenarios of data analysts at different levels in China are different.
Basic data analysts basically sort out data reports, write SQL statements and check data every day. No data analyst can skip this stage, and they need to start from the grassroots.
Middle level data analysts have a certain ability to work independently. In addition to doing some data report work, they will independently undertake some independent issues for special analysis, such as why sales will decline and how the current operation situation is, and then build a data indicator system to describe the current situation and analyze problems.
Senior data analysts, or department leaders/directors, are meeting almost every day. Meetings of management and other business departments will be held. They generally don't touch data, and most of their work is communication. Of course, in addition to meetings, they will also analyze problems and sort out decision-making suggestions from the perspective of senior management.
However, no matter what position you are in, data analysis is just the starting point, and using data to drive business to drive enterprise management is the real end point of value. Therefore, we should learn more about business. See how each department carries out its work, get familiar with the business process, look at the report, take the initiative to think about and find problems, and see how they translate problems into concrete measures. I believe that following this route, you will gain something on the road of data analysis.