1, theoretical requirements and sensitivity to numbers, including statistical knowledge, market research, model principles, etc.
2, the use of tools, including mining tools, databases, commonly used office software (excel, PPT, word, brain map), etc.
3. Ability to understand business and sensitivity to business. Have a deep understanding of business and products, because the starting point of data analysis is to solve business problems. Only by understanding the business problem can it be transformed into a data analysis problem to meet the requirements of the department.
4. Ability to express reports and charts. This is a key analysis model. If it is not well displayed to leaders and customers, its effectiveness will be greatly reduced, and it will also affect the career promotion of data analysts.
Second, please cultivate data analysis as a kind of ability.
Broadly speaking, most of the work now requires analytical ability, especially today when the concept of data operation is in-depth. Companies like BAT emphasize full participation in data operations. Therefore, treating it as a kind of ability training will benefit you for life.
Three, from the four steps of data analysis to see the ability and knowledge that data analysts need:
The four steps of data analysis (different from the process of data mining: business understanding, data understanding, data preparation, model establishment, model evaluation and model deployment) show the process of data analysis from a more macro perspective: obtaining data, processing data, analyzing data and presenting data.
(1) get data
The premise of obtaining data is the understanding of business problems. To turn a business problem into a data problem, we should discover the essence through the phenomenon, determine the latitude of analyzing the problem, and collect the data after clarifying the problem. This link requires data analysts to think and understand business problems in a structured way.
Recommended books: The Golden Pagoda Principle, McKinsey Trilogy: McKinsey Consciousness, McKinsey Tools, and McKinsey Method.
Tools: mind mapping, thinking management software.
(2) Processing data
A data analysis project usually takes up more than 70% of data processing time, so using advanced tools is conducive to improving efficiency, so try to learn the latest and most effective processing tools. The following are the most traditional but effective tools:
Excel: It is often used in daily notices, reports and sampling analysis. Its chart function is very powerful, and it can easily handle 654.38+ million-level data.
UltraEdit: Text tool is easier to use than TXT tool, and it is faster to open and run.
ACCESS: desktop database, mainly used for daily sampling analysis (it takes a lot of resources and time to do full-scale statistical analysis, and usually analysts will randomly select some data for analysis). Using SQL language, processing 654.38+0 million data is still very fast.
Orcle and SQL sever: These two types of databases are needed to process tens of millions of data.
Of course, if your ability and time permit, learning the recently popular distributed database and improving your programming ability will also be of great help to your future career development.
Analysis software mainly recommends:
SPSS series: the old statistical analysis software, SPSS Statistics (partial statistical function, market research) and SPSS Modeler (partial data mining), is easy to learn without programming.
SAS: The old classic mining software needs programming.
R: Open source software, a new and popular software, is more efficient in dealing with unstructured data and needs programming.
With the further development of text mining technology, there is an increasing demand for the analysis of unstructured data, and the use of text mining tools should also be further valued.
(3) data analysis
In order to analyze the data, we need to use various models, including association rules, clustering, classification, prediction models and so on. One of the most important viewpoints is comparison. Any data needs to be compared in the frame of reference before the conclusion is meaningful.
Recommended books:
1, Data Mining and Digital Operation, Ideas, Methods, Skills and Applications, Lu Hui, Machinery Press. This book is the best written by China in recent years. Be sure to read it as a Bible.
2. "Who says a rookie can't analyze data (introductory chapter)" and "Who says a rookie can't analyze data (tool chapter)", edited by Zhang Wenlin and others. It is an introductory book, suitable for beginners.
3. Statistics, fifth edition, edited by Jia Jun, Renmin University of China Press. A better statistical book.
4. The complete version of Introduction to Data Mining, with Pang Tan Ning waiting, translated by Fan Ming, People's Posts and Telecommunications Publishing House.
5. The Concept and Technology of Data Mining, translated by Han Jiawei and Fan Ming. , machinery industry press. This book is relatively difficult.
6. Quantitative Analysis Method and Application of Market Research, concise editor, Renmin University of China Press.
7. Practice of Questionnaire Statistical Analysis-Operation and Application of ——SPSS, Wu, Chongqing University Press. Well-known books in the field of market research explain the analysis of questionnaire data in detail.
(4) Presentation of data
This part needs to effectively present and report the data results, and needs to use the golden pyramid principle, charts, PPT and word to cultivate good speech skills.