For example, data analysis thinking, structured thinking, formulaic thinking, thinking of the learning method system ....... These mindsets help you, even when you come across a problem you are not familiar with, to be able to cut through the analysis from a certain angle and maintain clear logic;
A certain level of business comprehension, to be able to understand the business thinking behind the business. Only when you understand the problem can you convert it into a data analysis problem and know how to set analysis goals and conduct analysis;
Basic theoretical knowledge: mathematical statistics, modeling principles, recent market research, etc.;
The use of conventional analytical tools: commonly used office software (Excel, PPT, mind mapping), databases, statistical analysis tools, data mining, etc.;
Ability of data reporting and data visualization. If you can't analyze the data well enough, if you can't "express" it in a concise and easy-to-understand way, the effectiveness will be greatly reduced.
So what should we do to improve these abilities? The following specifically say how to do can put these basic strengths.
From the analysis of theory and tools to practice
1, analysis of theory
Theory of analysis includes: clear business scenarios, determine the goal of the analysis, build the analysis system and sort out the core indicators.
What we have to do is, first of all, to clarify what kind of business scenarios, different business, the analysis system is also different; and then, combined with the business problem to determine the goal of analysis, listed in the core indicators, and then collect and organize the required data.
Recommended books: "Data Management", "Battle of Big Data"
Data analysis of several steps:
(1) data acquisition
Data acquisition often seems simple, but it requires analysts to understand the business of the problem, that is, transformed into a data problem to be solved, such as, what data is needed, from which point of view to analyze, etc., after defining these issues, and then data acquisition. these problems, then data acquisition takes place.
This part of the process requires the data analyst to have structured logical thinking.
Recommended books: "Pyramid Principle", McKinsey trilogy: McKinsey awareness, tools, methods
Recommended tools: mind mapping tools (Xmind Baidu Brain Map, etc.)
(2) Data Processing
The data processing needs to master the efficiency of tools:
Excel and high-end skills:
Basic operation, function formulas, pivot data. Basic operations, function formulas, pivot tables, VBA program development.
I usually go through the basics first, know what's what, and then find a few cases to practice. More excelhome forum, usually think more about how to use excel to solve problems, make good use of plug-ins, and remember to save.
Professional reporting tools:
(into the scale of the enterprise will use) daily reports can be designed to do a general template, as long as you can write SQL can get started.
Compared to excel do report, this tool development of technical requirements are lower, can quickly develop routine reports, dynamic reports.
The use of databases:
Proficiency in SQL language (very important!!!). The common ones are Oracle, SQL sever, My SQL and so on.
Learn the popular hadoop and so on distributed database to enhance personal ability, job hunting and so on will be helpful.
(3) Analyzing data
Analysis of data often requires various types of statistical analysis models, such as association rules, clustering, classification, predictive models and so on.
Therefore, proficiency in some statistical analysis tools can not be avoided:
lPSS series: the old statistical analysis software, SPSS Statistics (biased statistical functions, market research), SPSS Modeler (biased data mining), without programming, easy to learn.
SAS: classic mining software, requires programming.
R: open source software, the new popular, more efficient in unstructured data processing, programming is required.
Various types of BI tools: Tableau, PowerBI, FineBI, for the processing of data can be freely visualized and analyzed, the chart effect is amazing.
Recommended books:
"Said the rookie will not be data analysis" series, entry-level books, beginners are most suitable.
"Data mining and data-based operations in practice, ideas, methods, techniques and applications", the content is very systematic and comprehensive.
"Market research quantitative analysis methods and applications", Jane Ming and other editors.
(4) data visualization
Many data analysis tools have covered the data visualization part, then only need to data results for effective presentation and speech reporting can be used word \ PPT \ H5 and other ways to show.
2, tools for practice
(1) for beginners, it is recommended to start with Excel tools, here to Excel as an example:
Learning Excel is a step-by-step process:
Basic: simple forms of data processing, printing, querying, filtering, sorting
Functions and formulas: commonly used functions, Advanced data calculations, array formulas, multi-dimensional references, functions
Visualization of charts: graphical presentation of graphs, advanced charts, chart plug-ins
Pivot tables, VBA program development ......
More excelhome forums, usually more thinking about how to use excel to solve problems, learn to use a variety of plug-ins, to be able to skillfully use Excel are helpful.
Functions and pivot tables are two key points.
Functions
The production of data templates must master the excel function:
Date function: day, month, year, date, today, weekday, weekknum Date function is to do the analysis of the template is necessary, you can use the date function to control the display of the data, query the data of the specified time period.
Math functions: product, rand, randbetween, round, sum, sumif, sumifs, sumproduct
Statistical functions: large, small, max, min, median, mode, rank, count. countif, countifs, average, averageif, averageifs statistical functions in data analysis has a pivotal role in the search for the average, maximum, median, median are used.
Find and reference functions: choose, match, index, indirect, column, row, vlookup, hlookup, lookup, offset, getpivotdata The role of these functions need not be said, especially vlookup, will not be the function basically Complex reports are difficult to move.
Text function: find, search, text, value, concatenate, left, right, mid, len These functions are mostly used in the data sorting stage.
Logic functions: and, or, false, true, if, iferror
(The above learn, basically can kill 90% of the office white-collar!)
Pivot table
The role of the pivot table is to generate a large amount of data can be interactive reports, it has some of these important features: classification and summary, take the average, the maximum and minimum, automatic sorting, automatic screening, automatic grouping; can be analyzed as a percentage of the year, year-on-year, the ring, the fixed ratio, custom formulas and so on
Realistically, the take the number or report + EXCEL + PPT still seems to be the mainstream form.
Tools, both business people and analysts, can be automated to take the number of tools or BI tools to create reports to reduce the time of repeated operations.
Secondly, increase communication with business people, fully understand the business needs, when your business level and their level is almost or even higher, naturally know what their real needs behind a word or two.
Finally, standing on a higher perspective, the basic granularity of the report is the indicator, you can sort out the basic indicators of the enterprise system, from the perspective of business analysis to do the report, the report of the standardization of work, reduce the redundancy of the report, to avoid moving to do a report. Standardization includes indicator classification, indicator naming, business caliber, technical caliber, realization and so on. In fact, the ultimate goal is to achieve consistency of report data, reduce duplication of report development, reduce system overhead strategic initiatives.
In your spare time, you can add more mathematical and statistical knowledge, learn R, Python language, learn commonly used mining models, to the development of senior analysts!
Cheer up together, ducks!
The above, is today's sharing, data analysis ability sounds very big and abstract, although it is soft strength but is the industry's hard requirements! Quantitative change causes qualitative change, step by step, in order to do the analogous, do up the project will be more and more smooth.