Big data analyst with annual salary of 500,000 make a note of "excerpt"

The following is a data analysis in the field after the roll of n years, write down some of the experience, will be able to give some reference to the new place. (summed up good, you can learn from learning OH)

What are the requirements of the data analyst?

1, the theoretical requirements and sensitivity to the numbers, including statistical knowledge, market research, model principles.

2, tool use, including mining tools, databases, common Office software (Excel, PPT, Word, brain map) and so on.

  3, business understanding ability and business sensitivity. To have a deep understanding of business and products, because The starting point of data analysis is to solve business problems, only to understand the business problem, can be converted into data analysis problems, so as to meet departmental requirements.

4, Report and chart presentation ability. This is a well-done analysis model, if not well presented to the leaders and customers, the results will be compromised, will also affect the data analyst career promotion.

Second, please use the data analysis as a kind of ability to cultivate

In a broad sense, most of the work now requires analytical capabilities, especially in today's data-based operations, where companies like bat emphasize full participation in data-based operations, so it will be a lifelong benefit to you as a competency training.

Third, from the data analysis of four steps to see the data analysts need to have the ability and knowledge:

The four Steps of data analysis (which differs from the data mining process: Business understanding, data understanding, data preparation, model building, model evaluation, model deployment), is the process of presenting data analysis more broadly: acquiring data, processing data, analyzing it, and presenting data.

(i) Access to data

The premise of acquiring data is to understand the business problem, to transform the business problem into a data problem, to find out the nature of the phenomenon, to determine the latitude to analyze the problem, to define the problem, and to. This link requires the data analyst to have structured thinking and understanding of business issues.

Recommended books: Pyramid Theory, McKinsey trilogy: McKinsey awareness, McKinsey tools, McKinsey methodology

Tools: Mind Mapping, MindManager software

(ii) Processing of data

A data analysis project, typically with data processing time of more than 70%, the use of advanced tools to improve efficiency, so as far as possible to learn the latest and most effective processing tools, the following is the most traditional, but very efficient tools:

Excel: Routinely used in notifications, reports, and sample analysis, its charting is powerful and it's easy to handle 100,000 levels of data.

UltraEdit: Text tools, easier to open and run faster than TXT tools.

Access: Desktop database, mainly for daily sampling analysis (to do full-scale statistical analysis, consumption of resources and time, often analysts will randomly extract some of the data for analysis), the use of SQL language, processing level 1 million data is still very fast.

Orcle, SQL Sever: These two types of databases are needed to work with tens data.

Of course, in their ability and time permitting, learning new and popular distributed database and improve their own programming ability, for the future career development will also be a great help.

Analysis Software main recommendation:

SPSS series: Veteran statistical analysis software, SPSS Statistics (partial statistical function, market research), SPSS Modeler (partial data mining), without programming, easy to learn.

SAS: Classic mining software, need programming.

R: Open source software, the new popular, for unstructured data processing efficiency higher, need programming.

With the further development of text mining technology, the demand for the analysis of unstructured data is increasing, and the use of text mining tools needs to be paid more attention.

(iii) Analysis of data

Analysis of data, need to use a variety of models, including association rules, clustering, classification, prediction model, one of the most important idea is the contrast, any data needs to be compared under the frame of reference, the conclusion is meaningful.

Recommended Books:

1, "Data mining and data operation actual combat, ideas, methods, techniques and Applications", Lu Hui, Mechanical Press. The book is the best written in the country in recent years, so be sure to read it as a Bible.

2, "Who said rookie will not data analysis (introductory article)" and "who said rookie will not data analysis (Tools)", Zhang Wenlin and so on. An entry-level book for beginners.

3, "Statistics" fifth edition, Jia Junping and so on, Renmin University of China Press. A relatively good book of statistics.

4, "Introduction to Data Mining," the full version, [Mei]pang-ning tan wait, Ming fan, such as translation, People's post and telecommunications press.

5, "Data mining concept and technology", Jiawei Han, such as Ming fan, such as translation, mechanical industry press. This book is relatively difficult.

6, "Market research Quantitative Analysis method and application", concise and so on, the People's University Press.

7.-SPSS operation and application of questionnaire statistical analysis practice, Wu Minglong, Chongqing University Press. In the field of market research is a more well-known book, the Questionnaire survey data analysis to explain more detailed.

  

(iv) Presentation of data

This part needs to make the data result effective presentation and presentation, need to use pyramid principle, chart and PPT, word presentation, cultivate good speech ability.

Recommended Books:

1, "Persuasive let your ppt will speak", Zhang Zhijin and so on, the people post and Telecommunications press.

2, "Don't tell me you understand ppt" reinforced version, Lizhi, Peking University Press.

3, "Speak with a chart", Keane. Zerazny, Ma Xiaolu and other translation, Tsinghua University Press.

(v) Other knowledge structures

In addition to having mathematical knowledge, data analysts also have the knowledge of market research, marketing management, psychology, behavioral science, product operation, Internet, big data, etc., and need to build a comprehensive knowledge system to support the different types of business problems encountered everyday.

Recommended Books:

1, "Consumer Behavior Science" 10th edition, Schiffman and so on, Jianglin translation, Renmin University of China Press, should now be updated to a higher version.

2, "Grotesque Behavior" upgrade version, Eric Li, Zhao Deilang, such as translation, Citic Press

3, "Marketing Management", Kotler, Meiqinghao translation, Gezhi Press and Shanghai Publishing House co-publishing

4, "Internet thinking-alone nine swords", Zhao Dawei editor, Mechanical Press

5, "Big Data era-the big change of life, work and thinking", Schoenberg, Zhou Tao and other translation, Zhejiang People's Publishing house

Iv. career development for data analysts:

1, data analysts usually divided into two categories, different division of labor, but each has an advantage.

One is to engage in data mining and analytical work within a dedicated mining team. If you can learn to grow in this kind of professional team, it is fortunate, but the threshold to enter such a team is high, need solid data mining knowledge, mining tool application experience and programming ability. Such analysts are more inclined to technical lines, and future career paths may take the technical route of experts.

The other is a data analyst who sinks to various business teams or operations to become part of a business team. They work to support business operations, including abnormal monitoring of daily operations, customer and market research, participation in product development, building data models to improve operational efficiencies, and more. This type of analyst favors products and operations and can move to operations and products.

2, data analysts the ideal industry in the Internet, but all the way through Rome, take the right route.

From an industry point of view:

1) The Internet industry is the most widely used data analysis industry, among which e-commerce enterprises, is currently the most fire, and enterprises also pay more attention to the value of data analysis, is the ideal growth platform for data analysts.

2) Second, consulting companies (such as the dedicated data mining company Teradata, Nielsen and other market research companies), they need data analysis talent, and relatively speaking, the data analysts in the consulting company growth faster, professional will be more comprehensive.

3) Again is the financial industry, such as banking and securities industries, the industry's reliance on data analysis needs, more and more.

4) Finally, the telecommunications industry (China Mobile, unicom and telecommunications), they have a huge amount of data, in the severe competition, but also more and more attention to data analysis, but the threshold of entry to these companies is relatively high.

Five, who is suitable for learning data analysis?

The answer to this question is the same as "who is fit to learn Kung Fu", no doubt that Kung fu is suitable for anyone to learn (excluding sinister designs) because he can keep fit. The effect of Kung Fu is to see the depth of practice of the martial arts practitioners. Often people argue, is Yong Chun Boxing, or sanda, in fact, reverse the cause and effect, should see which people practice better, there is no high or low level, only the thickness of the cultivation of people.

In fact, the subtext of the problem is "who learns data analysis and is more likely to succeed (such as career success)", depending on your interests, dedication and opportunities. But to be outstanding, you need a little talent in addition to the top three points, where opportunities are the career development platform, business environment, mentors and colleagues you meet.

Borrowing management guru Drucker's words "management can be learned", management is not natural, and the ability to analyze data can also be improved the day after tomorrow. You may be good, you need more effort + interest, and the process of this effort also includes the part where you look for opportunities.

Vi. about how to learn:

The most important way to learn is to find the right one, and it is best to learn from the problems you encounter with your work.

1, collect books, case library and video, first understand the theory, and then learn the software operation, own production of their own tutorials.

For example, you learn a cluster analysis model. 1) Collect relevant clustering analysis model books, cases and teaching videos, understand the principles of clustering analysis, the main types of algorithms (division, hierarchy, density, grid), the scope and premise of the model, how to evaluate the accuracy of the model.

2) Learn to use software to achieve.

3) summarize and organize into a PPT and production operation video, become their own learning tutorials, and constantly improve.

4) After learning to a certain extent, can be in the blog, and other channels to share, to give people and fishing, but also some of their own harvest.

2, pay attention to celebrities, famous Bo, website, multi-channel learning.

1) Focus on professional data analysis, consulting company web site and forum, especially stressed that the statistical software company's website, such as the Web site of SPSS has a lot of case library, it deserves attention.

SPSS Case Library, can search all kinds of cases on the official website: http://www.ibm.com/developerwork ... 8zhangzy/index.html

In addition, you'd better build your own Web site navigation directory, improve your learning efficiency

2) Pay attention to the celebrity Bo, it is best to add their Weibo, and the public number, look at the cattle People's blog and other content, or can get a lot of guidance, this you understand.

3) Join some QQ groups that share a common hobby and learn to communicate with each other. Usually people in the group will put forward some real operational problems, and then we use different methods to solve, the idea is very enlightening.

4) Learn from fragmentation to maximize your time value. In order to make use of the scattered time, usually I will upload some information to the network disk, in a fragmented time through the mobile phone for video, document learning and so on. Currently using Baidu Cloud disk and network disk. Baidu Cloud Disk application is relatively broad, usually in the network search "keyword + Baidu cloud", search results can be stored directly on the cloud disk, the speed of the searching save greatly improved. Network disk space is relatively large, can reach 40T, while there is a safe encryption function, high security.

Install some apps on your phone and learn anytime, anywhere.

VII. Final Recommendations

Please ask yourself again if you really like data analysis, can you endure the loneliness of processing data? If so, start learning and give you a few suggestions.

1, the data analysis as a kind of ability to cultivate, let oneself in the present team to show good data analysis ability, for you after the internal transfer ready. If the internal transfer does not work, you can consider moving to the industry I analyzed before, but I strongly suggest you still need to learn the programming ability of the system development, and the Business Intelligence System (BI and CRM) have a certain understanding, which may be the advantage of the application of data analysis. If there is no data analysis experience to apply, it will be more difficult, the employer's examination of your statistics and data mining model knowledge, as well as the use of tools.

2, in the company to find some colleagues have a common hobby to study data analysis, usually more data analysis to do a good colleague, it mountain stone, can attack Jade.

3, solid learning one, two data mining software, based on your programming basis, it is recommended that you can learn SAS or R, while supporting the study of SPSS Modeler. If you don't have a programming foundation or want to get results in the short term, you can learn SPSS first. SAS+SPSS, basically can meet the needs of a large number of enterprises, the three will, that better.

4, to understand how the company is operating, how the product is developed, how to do customer research to lock customer needs, how to do product marketing, these need to constantly work to accumulate and extensive reading.

5, start learning, first read a few interesting data analysis of the book, and then the system to learn statistical knowledge (the proposed textbook with the fifth edition of Statistics, Jia Junping, etc.), and then on-line rapid collection of software operation video and case, and then analyze the model to learn and summarize, Learning can best be done with the problems of the actual work.

6, to a certain extent, to participate in some data analysts professional certification, further carding the knowledge structure, at the same time know some like-minded friends and teachers, but also to you a great help.

Hope you can become the person you want to be!

Original: http://bbs.pinggu.org/thread-5045933-1-1.html

Big data analyst with annual salary of 500,000 make a note of "excerpt"