In the era of big data, data is king. Using data to do analysis and provide guidance for company decision-making is something that many companies must do in this era of refined operation, cost reduction and efficiency improvement. All major enterprises have established data analysis departments. Up to now, the domestic data analysis talent gap is140,000, and it is estimated that the market scale will reach 200 billion in 2025! Data analysis is not so much a job as an important skill. Having this skill means that your employment prospects are better and your career development is broader.
So how to learn zero-based data analysis? Next, I will give you a comprehensive understanding of the data analysis industry from three aspects: learning cycle, learning content and career development planning.
1. How long does it take to learn data analysis?
Everyone's learning ability and foundation are different, so the learning cycle of data analysis is different. If it is through self-study, this cycle may be very long, because there is no professional teacher's guidance and no systematic study. Generally speaking, if zero-based learners carry out systematic training, it will take nearly three or four months at the earliest. The study of data analysis should start with familiarity with tables and table structures, and its origin must be based on familiarity with Excel and the ability to extract numbers from the database in order to improve skills. Your skills range from being able to extract numbers from the database, processing tens of thousands of rows of small data with Excel and BI, to processing hundreds of thousands or even millions of rows of medium-sized data in batch with python, and finally processing tens of millions or even billions of big data with related components of big data, such as hadoop, spark, flume and so on. The tools and methods required for each stage are different. Generally speaking, for self-taught analysts who can handle medium data, they should at least know python's panda, numpy and other data processing libraries. This cycle of zero self-study is generally related to understanding and self-discipline. Students with high understanding and self-discipline may be able to master it in 4 months; If students don't have understanding and self-discipline, this cycle may give up halfway and it is impossible to estimate the time. The practical employment class of data analysis (Juju College) is recommended here, which focuses on cultivating the data processing ability, data analysis ability and data mining ability of data analysts. The course content includes database management, statistical theory and methods, the application of mainstream data analysis software and data mining algorithms. , and explained a set of data analysis process technology with actual combat practice system. After learning, learners can directly reach the level of data analysts.
2. What should I learn in data analysis?
( 1) Excel
Speaking of Excel, some people may think it is simple, but Excel is indeed a powerful weapon. The data analyst of zero basic science must start with Excel, because Excel is the most used tool to deal with small-scale data enterprises and plays an extremely important role in basic data analysts and data operations. As the core tool of data analyst, the specific learning content includes Excel function skills (search function, statistical function, logical function) and Excel fast processing skills (format adjustment, search positioning, shortcut key skills, etc. ) and Excel visualization skills (combination chart, bar chart, data bubble chart).
(2) Mysql
SQL is also the core content of zero-based learning data analysis. Because as a data analyst, the first problem you have to solve is that you have data to analyze. Usually, enterprises will have their own databases, and data analysts must first know what data they want to extract from the enterprise database according to business requirements. If the enterprise deploys a local database, the extracted data must be in SQL language. SQL is easy to understand and easy to use, so you must learn it. SQL language begins with learning MySQL database, which involves adding, deleting, modifying and querying table structure data. In real enterprises, data analysts generally have no right to add, delete and modify, only the right to check. Students should focus on mastering various sentence patterns.
(3) Python
The foundation of Python is very important for data analysts. For the data volume of100000 or1000000, both Excel and BI will be completely unusable because of the jam. However, in practical enterprise applications, it is very common to process 100,000-level and millions-level data at a time. Python is a powerful tool for processing this kind of middleweight data. Because Python has many powerful third-party libraries, such as Numpy, Pandas, Matplotlib, Seaborn and so on. These libraries enable data analysts to clean up and analyze millions of data. Python can not only clean data and draw pictures, but also analyze big data algorithms with sklearn. Although Python is an important tool for data analysis, it has different levels of mastery in different career development directions.
(4) BI business intelligence tools
BI can be understood as an advanced version of Excel chart pivot table. BI is to connect tables with each other and then draw many indicator diagrams. This is a big screen kanban, as shown below:
BI kanban diagram
Enterprise sales indicators, operational indicators, logistics indicators and so on. These graphs can show the average sales unit price in the past five months, the change curve of logistics shipments sold in the past 24 months, and even the real-time sales now. These are all issues that enterprises are concerned about. With this kanban, leaders have very intuitive data in monitoring the business of enterprises, so that they can make timely decision-making adjustments. At present, the more popular BI software on the market are FineBI, PowerBI and so on. These BI softwares are actually very similar and not difficult to learn. From getting started to mastering FineReport and FineBI, we can quickly tap the value of data, turn these data into useful information, and provide data basis for enterprise decision-making, thus driving enterprise decision-making and operation.
(5) Mathematical statistics and data operation
Mathematical statistics and data manipulation methodology are the theoretical cornerstones of data analysts. Mathematical statistics includes probability theory, statistics, linear algebra and basic calculus theory. These contents do not need to be deeply understood, but their principles and connotations need to be mastered. Because the source of the whole data analysis is actually based on descriptive statistical analysis. Descriptive statistical analysis is the total number and average value of statistical samples; The algorithm involved in data analysis is the deeper architecture modeling in statistics. Therefore, mastering the relevant knowledge of mathematical statistics is basic and necessary for entry-level data analysts.
What is the methodology of data operation? The methodology of data operation is actually to learn the analytical models of various industries' operations. For example, for e-commerce, funnel analysis can analyze the number of people entering the home page PV 1, the number of people entering the clothing sector PV2, PV2/PV 1 and get a proportion of people entering the clothing sector. There are also many general analysis models: correlation analysis, A/B test and so on. For the data analyst who wants to develop the management route, data operation is the knowledge that must be learned. In fact, the knowledge of data operation is not complicated, that is, according to their own business needs, the indicators are disassembled to the finest, and then two data analysis methods are used.
(6) machine learning
The last advanced level requires data analysts to master the ability to analyze a large amount of data. This kind of analysis not only describes statistical analysis and data operation methods, but also includes predictive analysis. The essence of prediction analysis is to use the existing data to calculate the relationship between a group of variables X and the predicted final value Y (that is, the mathematical algorithm formula), and then use this algorithm to input more X into the algorithm to get a predicted value Y. It doesn't matter if you don't understand it here. In short, the data analysis at this stage is to use a large number of historical data to construct a set of mathematical formulas (that is, algorithms) and use this mathematical formula to predict the future. For example, if a person brushes a lot of short sports videos, according to the algorithm, it can be concluded that this person may be interested in watching football matches by Tencent sports members. This kind of inference and prediction is of great significance to the business world. If you want to be a data analyst who can master algorithms, machine learning is an introduction that cannot be skipped. Students should learn from simple single regression, multiple regression and logistic regression, and gradually learn more advanced algorithms such as decision tree, random forest and SVM.
3. Career development planning of data analysis?
Generally speaking, data analysis has two routes, one is management route and the other is technology route. Development to management, such as junior data analysts, to data operations, to data analysis managers, data operations directors and so on. This development path mainly needs statistics, Excel, PPT and other skills, and needs to write a market analysis report. This road seems that the technical mastery is not too deep, but the understanding of the business is extremely profound. And a deep understanding of business needs time and a deep spirit of business research. If you are not a friend of mathematics, computer and statistics, it is more suitable for this non-technical career development path.
And to develop in the direction of technology, the goal will be clear. One is to develop in the direction of data mining and learn cutting-edge algorithms such as deep neural network and NLP. The second is to go deep into data analysis and development and learn big data components such as hadoop and spark. This is a development direction of technology, which requires higher statistical ability, mathematical ability and programming ability.
In fact, both non-technical business direction and technical expert direction need two words: research. Of course, after hearing this, we also need to elaborate: it is not difficult to get started with primary data analysis. It is difficult to become an excellent data analyst in the second half, and you need to study hard.
If you see this, you feel that you are psychologically prepared for the entry direction of data analysts, but if you have no foundation, you really don't know how to enter the industry. You are welcome to chat privately, get the outline of data analysts' knowledge points for free, and do the entry consultation for data analysts for free.