The 11th CDA Data Analyst Certification Exam ended successfully on December 28, 2019. This exam set up 37 test centers in 23 cities across the country.

Recently, we interviewed several excellent students who ranked among the best in this exam, and in this article we have compiled their test preparation and learning experience. I hope that test preparers can have some reference and achieve ideal results.

Today I bring you a few candidates who have achieved excellent results in the CDA certification exam Level II. Let's take a look at their style below!

LEVEL II modeling analyst

First place Duan Xiaoli

1. Current work

Currently working in a joint-stock bank, engaged in the research and development of algorithm models, responsible for the support of algorithm models in the fields of risk management, marketing, customer service and so on.

2. Opportunity to apply for CDA certification exam

Thank you very much for my company. In 2019, the company carried out machine learning training for employees. In order to ensure the training effect, the company organized internal testing after the training and selected 10 employees to participate in the CDA level 2 certification exam. I was fortunate to be a member of the CDA certification exam.

3. How did I prepare for the exam

I prepared three or four weeks, mainly at night and on weekends. When preparing for the exam, I used a small strategy. The multiple-choice questions had very few scores. The test scores depended on the performance of the case operation questions. Therefore, for the preparation of the multiple-choice questions, I mainly looked at the exam outline. Proficiency, as for the other knowledge points are usually accumulated. Case operation questions carried out the whole process of simulation operation on the two sets of simulation questions in the preparation materials, and the code was continuously optimized afterwards.

4. What are the knowledge and difficulties in the preparation

The difficulty is that there are some confusing options or words in the objective questions. If they are not distinguished, it is easy to make mistakes; there are some algorithms that have not been understood before, and it is difficult to have a deep memory in a short time; there are many missing values in the case operation questions , You need to use the appropriate missing value filling method.

5. Recommended books and courses

During the test preparation period, I mainly look at the 2019 training materials, videos and CDA level 2 certification exam outline. After the test, I decided to turn over the "Data Mining Technology" I saw before, and I should have some new understanding.

6. Suggestions for candidates

The exam involves a lot of content and a wide range. When preparing, focus on the key points. In addition, case operation questions must first understand the data and the business logic behind the data. Do not directly train the model as soon as you come up.

6. Future career development plan

In the future, I will learn to apply it, combine theory with actual business, try algorithm models in different scenarios, let the data sound, and create value.

Second place Feng Junqi

1. Opportunity to apply for CDA certification exam

I am a graduate student in statistics at Nanjing University of Science and Technology, and the direction is data mining. In recent years, data analysis and data mining are very popular. The CDA Level 2 modeling analyst's exam also combines machine learning algorithms and software operations. The instructor also recommends the exam, so apply.

2. How did I prepare for the exam

The test preparation starts in October, about two hours a day.

The test preparation is divided into two parts: theory and practice. The preparation time for the theory part is relatively long.

The theoretical part is to look at the outline, review it according to the outline requirements, and then study the details for each algorithm. Finally, do a simulation question, find your own shortcomings, and then check the shortcomings.

The practical part has a foundation to try the application of various algorithms, and find a case to operate. There is no foundation to learn basic knowledge first.

3. What are the knowledge and difficulties in the preparation

Some of the algorithm details in the theoretical questions are not easy to understand and require repeated thinking from multiple angles. Handling of unbalanced samples of operation questions, feature engineering, model tuning, etc.

4. Recommended books and courses

The book is mainly based on the exam outline. More details recommend "Data Mining: Concepts and Technology". This book is closely related to the outline and the content is very substantial; if you have the ability, you can look at the watermelon book, statistical learning methods, and the exam is not It will involve so deep, you can understand if you are interested.

The course is mainly in operating software, you can refer to some CDA official courses.

5. Suggestions for candidates

1) The contents of the outline should be mastered, and the reference book should be read as much as possible

2) If the review is in place, there is not much difference in the scores of theoretical questions. The focus is on practical questions, more hands-on and more attempts.

3) Do a good job in daily study and supplement yourself in free time

6. Future career development plan

Deepen the understanding of the model learned, pay attention to the preface model, strengthen the analysis ability, code ability, improve the proficiency of theory and operation, and pay attention to the summary.

Tied for third Huang Jianchuang

1. Current work

From 2011 to now is the ninth work working years, he has been working in the actuarial department of an insurance company, now mainly responsible for the quantitative analysis of assets and liabilities related work.

2. Opportunity to apply for CDA exam

In recent years, with the development of big data, artificial intelligence and other technologies, traditional actuarial technology will face more and more challenges in the future. On the one hand , applying for CDA wants to ensure its professional value by expanding its own knowledge boundaries; Learning algorithm-related content can provide new ideas and methods for company management decisions.

3. How did I prepare for the exam

First of all, according to your actual situation, you can count the time available for review every day: Monday to Friday, because you have to go to work during the day, you mainly use the free time at night to review, and insist on reviewing for 1 hour a day. The time for preparing for the review is relatively fragmented); weekends are more ample, and the review is more systematic, generally 2 hours each in the morning, middle and evening.

Then, according to the exam outline, the review time is allocated to the content of each chapter to formulate a review plan. For example, I already have a certain knowledge base for Bayesian statistics, regression, neural networks and other chapters, and less time allocation, while for relatively unfamiliar chapters such as random forests and integrated learning, I have more time allocation.

The last is to strictly implement the plan according to the review plan. In order to ensure the quality and progress of the review, it is inevitable to sacrifice the time for communication activities with relatives and friends during the preparation period.

4. What are the knowledge and difficulties in the preparation

When doing case operation problems, there are big problems, such as how to choose the appropriate algorithm. After the algorithm is selected, how to adjust the optimal parameters to improve the accuracy of model prediction or classification. If there are partners who can discuss the exam together, it will greatly reduce the trouble in this regard.

5. Recommended books and courses

The books recommended by the exam outline are all classic, and it is almost the same as selecting one or two of them according to the bibliography . If you want to quickly and effectively improve your professional level in data analysis, consider taking CDA-related training courses.

6. Suggestions for candidates

In addition to mastering various theoretical algorithms, the purpose of the modeling analyst exam is more important to apply the algorithm to practice, so the usual learning process is essential for case operations while reading textbooks, only through a large number of case programming Analysis can make you perfect, and further consolidate your understanding of various algorithms.

7. Future career development plan

In the future, we will continue to move forward on this actuarial path, hoping to integrate the knowledge learned by CDA with actuarial theory, and truly apply it to our daily work to more effectively reflect the value created by the company.

Tied for the third Lei

1. Current work

Originally engaged in data analysis work in Shanghai Telecom, this year just transferred to a data mining engineer and project manager.

2. Opportunity to apply for CDA exam

I have been using a relatively primitive method ( excel and other traditional tools) to do simple descriptive statistical analysis, so I hope to improve my data analysis capabilities. In May last year, I compared the better data analysis certifications on the market. The CDA was more in line with my needs, so I applied for CDA Level1. After a month of hard work, although statistics have almost no foundation, they passed smoothly. I also compared Level2 modeling and big data, and felt that modeling is more suitable for myself now, so I signed up for Level2 modeling immediately in September last year.

3. How did I prepare for the exam

Although I made up my mind to register for the exam in September last year, I started to learn about data mining immediately after I passed the CDA LEVEL1 test in June last year. At the same time, I have learned Python by myself, so there are still some basics.

From September to December before the exam, that is, a full four months, you need to work during the day, mostly in work leisure or evening time to study, and the average daily time spent studying is 3 hours. The main learning content includes:

1) Reading and comprehension of outline analysis

I have read the outline 4 times in total , once a month, every time I have a new experience.

The first reading, let me know which of my foundations are not correct and targeted adjustments.

Reading the second time, sorted out the mind map.

The third time I read it, I read it in combination with the next two simulated papers, and took notes on my notebook.

Read it for the fourth time, check and fill in the gaps, and finally review.

2) Self-study of simulation paper and official test bank

The simulation paper is very important. Many of the actual exams are based on the outline analysis and the original questions in the simulation paper. There may be slight changes, but as long as the score is clear, it will be easy.

3) Participate in a data mining competition

It is possible to use Kaggle and CDA practice games, but it is best to participate in an actual game. It will be more feelable to apply what you have learned in the game.

There are many related competitions, such as Kaggle, Tianchi, CCF, and even CDA's own competitions are possible.

4) Self-organization of knowledge points

If you don't do the finishing after learning, you will eventually forget it slowly. Before preparing for the exam, I spent a week sorting out what I learned and sharing it in the form of a blog, mainly including:

"Study on the characteristics, advantages and disadvantages of seven commonly used supervised prediction models"

" CDA LEVEL2 Outline Analysis Case Question Python Implementation Code"

" Python: Implementation of 3 Common Data Inspection Codes"

" Python: 14 common data cleaning codes"

" CDA Level2 Simulation Question 1 Python Code Implementation"

" CDA Level2 Simulation Question 2 Python Code Implementation"

4. What are the knowledge and difficulties in the preparation

CDA2 modeling is more focused on actual combat than CDA1, so it is more suitable for people like me who have actual combat greater than theory. What impressed me most was that when I was doing the second set of simulation questions, I encountered a problem of calculating Bayes. The calculated answer was inconsistent with the standard answer. The group discussed for a long time. Finally, I relied on the CDA teacher to give the idea of understanding the question. Therefore, group discussion is a good learning method, and only communication can make rapid progress.

5. Recommended books and courses

First of all, the CDA syllabus is the best review material. You can master at least 60? Knowledge points, and "Introduction to Data Mining (full version)" can basically cover the theoretical knowledge of 95?

Then for data mining, the tools used are generally Python, so there are 4 books worth reading: "Python Basic Tutorial (3rd Edition)", "Data Analysis Using Python", "Python Machine Learning Basic Tutorial", " "Machine learning combat".

For the final video and course, Wu Enda's " Machine Learning ", Tang Yudi's " Python Data Analysis and Machine Learning Combat" and the CDA official website video course are all good choices.

6. Suggestions for candidates

Modeling or data mining is a hot industry at present. It is easy to get started but it is very difficult to go deep. It requires a lot of practical experience and a good mathematical foundation. If you only use models and parameter adjustments, the path is not It will be too far, so try to delve deeper in your study and understand the principles behind the model and code, which will be of great help for future practical work.

7. Future career development plan

At present, I have just transferred to a new position, not only to undertake the management work, but also to be responsible for the research of data mining. At the same time, I also need to learn a variety of new knowledge such as network, cloud, and big data, so I feel painful and happy. I hope that I can use technology as the core in the past few years to improve my knowledge system and comprehensively improve my ability.

LEVEL II Big Data Analyst

First place Wu Wentao

1. Current work

I am currently working as a big data engineer in the PayPal project, mainly engaged in big data analysis and product development for mobile payment platforms

2. Opportunity to apply for CDA exam

Today's workplace is fiercely competitive, and holding a skill certificate is also one of the ways to improve your own competitiveness. At the same time, taking appropriate exams is also a test and review of your learning results. Therefore, you applied for the CDA at the end of the year .

3. How did I prepare for the exam

In fact, I am a person who is easily addicted to something, so I did not think too much in preparing for the exam. Basically, I devote all my time to accomplishing this goal; that being said, in order to study efficiently, I still have to find some methodologies. It took me about two months to prepare for the exam. Most of the time is not spent on learning new knowledge, but repeatedly reviewing the old knowledge before, the more difficult the knowledge points, the The more you need to understand them, and after reviewing them, immediately write them into a blog post, so that you can convert a short-term memory into a long-term memory, which is not easy to forget. American physicist Feynman also proposed a methodology for learning, that is Try to teach your knowledge to someone who does not understand it at all. If he can understand it, it means that you have really learned it.

4. What are the knowledge and difficulties in the preparation

Because I know that there will be parts on the computer involved in the exam, when I review the theory on one side, I pay more attention to the practical part, how to build a cluster environment, how to quickly locate the problem and find a solution, these are no shortcuts to find Yes, the only way is to say that the old man who sells You Wengli, "No other, only familiarity", only by repeatedly tossing and stepping on the pit, can you really exercise to face the pressure in the actual project, The ability to think and solve problems independently.

5. Recommended books and courses

You can combine both video tutorials and books to learn. The video recommends directly purchasing the training system of some training institutions to learn comprehensively, such as the nine chapter algorithm, etc. These courses are often very practical and fit the actual production environment of the project, but If you want to do a more in-depth study of a certain field, I recommend buying some books to calm down and nibble. For example, when learning the knowledge of ML, I bought "Machine Learning Combat" and "Machine Learning". "Linear Algebra Basics" and other books, and when learning the big data framework, it is to find the technical documents found on the Apache official website.

6. Suggestions for candidates

The exam is not the ultimate goal. The main function of the exam is to prepare for a better entry into the data industry. If you really want to start learning, then the CDA course is a good choice. The current mainstream programming languages such as Python, analysis Engines such as Spark will be introduced in detail in the course. For those who want to prepare for the exam, it is recommended to take enough time every day to study systematically, and be sure to stick to it, and believe that there will be results in the end.

7. Future career development plan

Industry: My vision for the entire industry is that in the future, the trend of digitalization and informatization in society will surely develop better and better, and technology will play an increasingly important role, bringing the backwaters to life, eliminating barriers to the industry. Things that take a long time to build are completed quickly. In the future , the development of AI, cloud, and big data will never be the only one on the Internet, but will become a sparse and ordinary thing like air, so be prepared in advance. Looking forward to the development trend of the society in the future, it is necessary to have some skills that may be used in the future.

Occupation: For practical business scenarios, how to really help traditional industries to do digital transformation, how to build a data center in order to truly respond quickly and support decision-making, these are some of the future career expectations; and in the future, the use of AI It will also become more and more convenient. Maybe everyone can build a set of machine learning frameworks with extremely complex underlying principles at a small learning cost. Everyone can be an engineer, and everyone can be a product manager. The threshold will be lowered, as long as there is a good enough idea, there is a way to achieve it.

Second place Cui Yupeng

1. Opportunity to apply for CDA exam

I am currently a graduate student of Traffic Information Engineering and Control Department of Beijing Jiaotong University.

I did a stream data analysis about Storm in my undergraduate program . During the graduate period, I became interested in the wider application of Spark. I hope that the systematic study and practical projects will be applied in future work. After comparing the big data related exams on the Internet, I found that the CDA exam syllabus is more systematic and reasonable, which is helpful for my systematic study and testing within a limited period, so I decided to apply for the exam.

2. How did I prepare for the exam

Three months of preparation for the exam , 3-4 hours of study per day, corresponding to each big data tool, borrow 1-2 books, and learn against the syllabus and books.

The first month: review Linux, build Hadoop and Spark clusters, learn the principles of Hadoop and Spark, and learn Scala programming.

The second month: use Spark to write examples, learn the principles of MySQL, Hive, HBase and the combined use with Spark.

The third month: integrate the knowledge learned, and compile the project based on the online combat tutorial to deepen the understanding of various big data tools. And according to the syllabus analysis for extended learning, understanding and memory.

3. What are the knowledge and difficulties in the preparation

1) The operating mechanism of Hadoop and Spark is not easy to understand. If you have the conditions, you should go to the library to find related books, read more and think more, and read the source code and debug the breakpoints to help understand.

2) SparkMLlib has more machine learning content, which is also the focus of practical operation. It should be combined with examples to deepen the understanding of each algorithm.

4. Recommended books and courses

"Bird's Linux Private Kitchen" is a relatively vivid book for Linux learning.

" Spark Programming Basics" is a good book to learn Spark.

" Hadoop Expert: Management, Tuning and Spark" is a good book for advanced learning of Spark and Hadoop.

5. Suggestions for candidates

1) As the big data ecology involves many architectures, students with no foundation should focus on learning with Spark, students with foundation should focus on combining Spark with various ecosystems, and learn or review related knowledge points through the examination system. Learning is helpful to read Spark source code and deepen the understanding of Spark principles and applications.

2) The content of the syllabus analysis is limited. It is necessary to organize the notes against the syllabus. The following is a summary of some of my notes.

3) The purpose of learning is to apply, not just exams, but every chapter should look for related exercises and do it manually, so that each part of the code is coded at least three times.

7. Future career development plan

More skilled application of big data ecology to achieve efficient and high-value data analysis and more accurate data recommendation.