Data analysis is a widely used skill label. In a real work environment, there are at least three types of positions that can be considered related to data analysts: BI (Business Intelligence), QA (Quantitative Analyst / Data Scientist), and BA (Business Analyst).
a) BA first defines the measurement method of the business, such as paying users or active users, and whether 1-day daily activity is important or 30-day daily activity is important. Then collect data for analysis of various dimensions, such as region, channel, user behavior and so on.
b) Based on QA and BA analysis, build an analysis model, which may use various statistical, optimization, and machine learning methods. The model has been evaluated by A/B testing, stability, etc., and the expected impact.
c) BI is responsible for automatic calculation, regular automatic update and visual display of all user metrics, market trends, model performance, etc. defined by BA and QA through database and SQL, and becomes the core data that the department pays attention to every day.
It can be seen from this that BA/QA/BI have different focuses: Business Analyst defines problems, analyzes data, proposes and partially implements solutions; Quantitative Analyst defines models and architectures to implement complete solutions; Business Intelligence Visualize, automate calculations, and update data and metrics.
What abilities do they need?
Back to the basics, whether it is BI, QA, or BA, basic data analysis skills are required. The difference is that the follow-up technology and business focus are slightly different. The following provides an advanced route for data analysis.
Data Analysis Learning Route
1、Excel
Most people have been exposed to Excel. The focus is to understand various functions, such as sum, count, sumif, countif, find, if, left/right, time conversion, etc.; Just search and use it whenever you need it. In addition, vlookup and pivot table are two cost-effective techniques. After these two are done, it is generally not difficult to count data within 100,000 pieces.
2. Data visualization
数据分析界有一句经典名言，字不如表，表不如图。数据分析的最终都是要兜售自己的观点和结论的，兜售的最好方式就是做出观点清晰数据详实的PPT给老板看。
虽然Excel也可以完成很多的数据可视化功能，但是如果想要得到更专业的可视化效果，还是建议学些编程方面的知识。
这里推荐微软的Power BI（商业智能）和帆软的FineBI。BI和图表的区别在于BI擅长交互和报表，适合解释已经发生和正在发生的数据。Power BI适合个人学习，FineBI适合企业级的应用。
Power BI
3、数据库
Excel对十万条以内的数据处理起来没有问题，但是互联网行业就是不缺数据。但凡产品有一点规模，数据都是百万起，这时候就需要学习数据库。SQL是数据分析的核心技能之一，从Excel到SQL绝对是数据处理效率的一大进步。
除了最基本的增删改查、索引、约束外，主要了解where，group by，order by，having， like，count，sum， min，max， distinct，if，join，left join，limit，and和or的逻辑，时间转换函数等。如果想要跟进一步，可以学习row_number，substr，convert，contact等。再有点追求，就去了解Explain优化，了解SQL的工作原理，了解数据类型和IO。
4、R/Python语言
是否具备编程能力，是初级数据分析和高级数据分析的风水岭。数据挖掘，爬虫，可视化报表都需要用到编程能力。而数据分析领域推荐使用的两种语言绝对是R和Python了，二者在数据分析领域的地位可以说是旗鼓相当，各有优势。
The advantages of R are written by statisticians. If it is a priori argument for calling, plotting, and analyzing various statistical functions, R undoubtedly has advantages. To learn R, you need to understand data structure (matrix, array, data.frame, list, etc.), data reading, graphic drawing (ggplot2), data manipulation, statistical functions (mean, median, sd, var, scale, etc.); development environment Rstudio is recommended.
Python is an all-purpose glue language with strong applicability and many branches. We focus on data analysis. Need to understand the calling package, function, data type (list, tuple, dict), conditional judgment, iteration, etc. Anaconda is recommended for the development environment.
5. Statistical knowledge
Statistics is the foundation of data analysis. It will take some time to acquire knowledge of descriptive statistics, including: concepts such as mean, median, standard deviation, variance, probability, hypothesis testing, significance, population, and sampling.
6. Analytical thinking
Good data analysis must first have structured thinking, which is commonly known as pyramid thinking. Mind map is a must-have tool; then learn about SMART, 5W2H, SWOT, 4P theory, six thinking hats and other frameworks. The analysis also has a framework and methodology, mainly around three main points:
1) A business cannot grow and analyze without indicators;
2) A good indicator should be a ratio or proportion;
3) A good analysis should contrast or correlate.
7. Business knowledge (user behavior, products, operations)
For data analysts, the business is actually more important than understanding the data methodology. But unfortunately, there is no shortcut to business learning, you must rely on a little accumulation in a certain industry.