Current location - Training Enrollment Network - Mathematics courses - How to write a data-intensive case study
How to write a data-intensive case study
Teach you how to write a data analysis report! There is a core case study!

Submarine sounding data analysis

Maybe you will encounter the following scenarios:

Scenario 1: As an operator, the boss will ask you the operation data of the month at the end of each month. Weibo and Xiaohongshu are the main channels of the company. In the case of similar investment ratio, the exposure and conversion rate of Xiaohongshu is twice that of Weibo, while the data of other competing companies are comparable. You don't want to give up the channel of Weibo, you need to persuade your boss to present a logical analysis report.

Scenario 2: As a sales supervisor, the overall sales of the company decreased by 20% compared with last year due to the epidemic, but your friend's company is also a sales company, and its performance has not decreased, but has increased by 15%. Through consultation, you found that their company changed the sales of most products to online, and also reached cooperation with several well-known anchors to make them famous on various platforms. At this time, you want to try to use this method to improve the company's sales performance. Years of experience in the workplace tell you that you need a detailed data analysis report to reach an understanding with the team.

Scenario 3: As a fresh graduate, you have entered a company internship that you have been interested in for a long time. If you behave well, you can stay and become a regular employee, but at the same time, the strength of the small partners who come in is not weaker than yours! At work, you find that your leader attaches great importance to the data thinking of subordinates and hopes that all reports can be combined with data. He thinks that all work reports lacking logic and data support are empty talk and hooliganism! At this time, if you can prove your logical thinking ability based on data to the leader and show it in the work report, the opportunity to become a regular employee basically belongs to you!

After reading these three scenes, maybe you will find out! Data analysis permeates all aspects of life and work. Whether it is upward reporting, downward management or enhancing the competitiveness of the workplace, it is necessary to master the ability of data analysis and make a well-founded and logically clear analysis report!

The purpose of data analysis report is to show readers valuable information such as analysis conclusions and feasible suggestions obtained in the process of data analysis, so that readers can have a correct understanding and judgment of the results and make targeted and executable strategic decisions according to the analysis conclusions.

The function of data analysis report is to analyze the analysis process-show the analysis results-and provide decision-making reference.

Seeing that there may be small partners here, there will be doubts. "What is a data analysis report?" "What should the data analysis report include?" "Is there anything to pay attention to when writing a report?" "Do you have any writing ideas?" . Don't worry, I will answer them one by one with years of data analysis experience!

I. What is a data analysis report?

Teams need to share and communicate, and data analysts need to gain insight into the data, share the analysis results with enterprise leaders, team colleagues, mass media and more stakeholders, and conduct all-round scientific analysis of project data through data analysis reports to evaluate the feasibility of the project, thus providing scientific and rigorous basis and reducing the risk of project investment.

Data analysis report is an important basis for judging the feasibility of the project. Any booming enterprise is based on the development of high-quality projects.

There are two kinds of data analysis reports, one is tracking analysis report and the other is research analysis report.

Tracking analysis report: The key to high-frequency presentation of daily business data lies in finding problems, not solving them. Generally used to answer "what's the matter". This kind of report often describes the operating conditions and finds problems through data, such as weekly reports and industry status analysis reports.

Research analysis report: used to answer "why", "why" and "how to do it". Such reports are usually used to solve specific business problems and provide effective solutions according to data analysis results, such as frequently asked questions diagnosis reports and decision-making suggestions reports.

2. Four things you need to know before writing a report

1. Understand the components of data analysis report.

Image source: BDA data analysis course

2. Clear what is a good or bad data analysis report.

A good data analysis report includes the following contents:

* * * consists of five parts: abstract, keywords, table of contents, text (including title, introduction, literature review, research process, conclusions and suggestions) and references (not less than 5 articles).

Like this! (The picture below is an excerpt from the report. )

Image source: BDA data analysis course

And meet the following data analysis requirements:

The research method requires:

Combination of quantitative analysis and qualitative research

Analysis process requirements:

Embodiment: data collection → data processing → data analysis → data visualization.

Suggestions on analysis methods:

Application: Comparative analysis, grouping analysis, cross analysis, regression analysis and other methods (not limited to the above analysis methods) to analyze the requirements of analytical tools;

A good report generally needs to include at least descriptive analysis and diagnostic analysis, that is, according to the target to be analyzed, at least an assessment of the current situation and a diagnosis of the problem should be given, and then a logical corresponding plan should be given.

Judging whether it is a good report mainly depends on whether the analysis logic is reasonable, whether the pictures and texts echo, whether the content is clear and easy to understand, whether the decision-making suggestions given can be implemented, and so on!

3. Determine the analysis industry and objectives

At work, maybe the boss or employer gave us a goal of data analysis, and we just need to solve the problem according to the established goal.

For example, the order volume of a product in the company has dropped by 20%. The boss hopes that you can find out the reason for the decline and give an executable plan! This is a clear data analysis goal!

It may be due to the recent lack of platform concessions, or other competing products have more advantages in price. What is the reason? All these require you to make a preliminary analysis.

In addition to the above scenes, I mainly introduce to my friends how to find the data analysis industry and target in their own projects.

1. Find your own interests.

Interest is the best teacher. With interest, we have the motivation to move forward and the impulse and desire to analyze and explore.

2. Find familiar industries and enterprises for analysis.

Choosing a familiar direction saves more time and experience than choosing an unfamiliar direction. You can have more time to find suitable data and data cleaning, analysis and exploration, and avoid putting too much experience on background understanding and business analysis.

3. Find the direction that you are good at or have resources.

Here are three common industries and three data analysis directions for your reference!

Image source: BDA data analysis course

Finally, I will give you some analysis directions related to your work:

Doing financial work can learn the data analysis of financial direction.

Doing operation and maintenance can do data analysis of human resource management.

Doing sales work can analyze target customers and sales volume.

After determining the analysis industry, it is necessary to clarify the objectives of data analysis. Here, I will introduce you from the two dimensions of growth and reduction!

Growth dimension: analyze the growth of income and efficiency.

Dimension reduction: analyzing the reduction of cost or risk.

In short, after defining the target industry, by comparing the past and present of the target industry, starting from the conflict between ideal and reality, we can find ways to increase income, improve efficiency, reduce costs or control risks.

Find the right data

Once the industry to be analyzed and the analysis target are defined, data can be collected. When collecting data, there is a common problem: how to find the right data for analysis? Some partners may ask, "I don't know. Can reptiles find suitable data for analysis? "

The answer is yes! There are two situations here, one is to find data from one's own work, and the other is to obtain data from open information sources.

1. Find data from your own work.

First of all, you need to think about whether your work needs improvement, which can be used as an analysis goal. Then go and see if there is any data to quantify. Sometimes data is not easy to obtain, such as from the company's new system or built-in database. You can also collect data, such as through questionnaires, which is a channel or way to collect data.

The benefits of finding data through work are: first, better understanding of background knowledge; second, finding the improvement points of work through data, so that the results of data analysis projects can also be reported to the boss, thus enhancing the competitiveness of the workplace and getting the boss's attention.

2. Looking for data from the Internet

If you can't find the data at work, you can try to find it online. There are three ways:

The first one is to obtain public data sets from online data competition platforms and data analysis communities.

Such as: Kaggle, Kosai (and Whale), Alibaba Cloud Tianchi, etc.

The second type: search through a data search engine.

Such as: Google data set search

The third kind: reptiles

Crawler through programming languages: Python, R language, etc. Through the stupid crawler tools: Houyi, Agaricus blazei and other tools.

The fourth type: common forum search.

Such as github, csdn, etc.

What kind of data is more conducive to analysis? It is suggested that partners can choose from the following four dimensions:

Image source: BDA data analysis course

Finally, there is another dimension. You can think about whether there are suitable business indicators that can be split according to business background. First of all, it depends on whether the underlying indicator data is mastered. If there is, we can analyze it by splitting the data indicators, and then find the corresponding required data.

For example: a data set about the sales of Tmall Double Eleven beauty products.

By splitting the sales indicators, we can see which factors affect the total sales volume and unit price, so as to find the sales law of beauty products and then put forward corresponding sales suggestions. This is the overall idea.

Then, the factors that affect the sales volume and unit price may be different brands and the evaluation of products, so one is to look at the quality and reputation of products through the platform quantity, and the other is the products of different categories, such as cosmetics, skin care products or other tools, which category is more popular with market consumers, which is a general analysis idea. After finding these data, we can start the analysis.

Third, how to write a data analysis report

1. Title

Title writing should be direct, accurate and concise, and strive to be fresh, vivid, unique and artistic. A good title can not only stimulate readers' interest in reading, but also reflect the theme of data analysis.

Commonly used title types are:

A. Summarize the main contents: focus on describing the basic facts reflected by the data, so that readers can see the key points of the report at a glance, such as "the order volume of XXX company increased by 65,438+05% compared with last year" and "the company's business grew rapidly in 2022";

B. Expounding the basic viewpoints: express and point out the basic viewpoints of the data analysis report with opinion sentences, such as "the retention of potential buyers cannot be ignored" and "lipstick products are an important pillar of the company's development";

C. Ask questions: ask questions analyzed in the report to arouse readers' attention and thinking, such as "What is the reason for the decline in order volume" and "Where is the company's development plan for the next three years";

D. Explain the analysis topic: reflect the analysis object, scope, time and content, but do not point out the analyst's views and opinions, such as the company's business development path in 2022 and the comparative analysis of departmental business;

2. Content

The table of contents is equivalent to the data analysis outline, which can reflect the analysis idea of the report. Catalogs can help readers find what they need conveniently and quickly. Therefore, the table of contents should list the names of the main chapters of the report and the corresponding page numbers. You can also list the more important secondary directories.

Some readers don't have time to read the whole report, but are only interested in some analysis conclusions shown in the chart. When there are not many charts in the written report, we can consider making the charts of each chapter into a separate directory for more effective use in the future.

3. Summary

Abstract is an overview of the report content, that is, an introduction to the report content. Abstract is a short article that concisely and accurately describes the important contents of the document, without comments and supplementary explanations. Its basic elements include research objectives, methods, results and conclusions. Specifically, it is the main object and scope of research work, the means and methods adopted, the achievements and important conclusions obtained, and sometimes other important information with intelligence value.

Pay attention to the following eight points when writing an abstract:

A. Contents that have become common sense in this field should be excluded from the abstract; Never write what appears in the introduction into the abstract; Generally don't interpret and comment on the content of the paper (especially self-evaluation).

B. don't simply repeat the information already in the title. For example, if the title of an article is "Study on Rhizome Formation in Vitro Culture of Several Orchids from China", then don't write at the beginning of the abstract: "For.

C. rigorous structure, concise expression and exact semantics. What to write first, then what to write, should be arranged in logical order. Sentences should be coherent and echo each other. Use long sentences with caution and keep them as simple as possible. Every sentence should be clear, and there should be no vague, general or vague words, but the abstract is a complete essay after all, and telegraph writing is not enough. The abstract is not segmented.

D. use the third person. It is suggested that the description methods such as "research …", "report …" and "investigation …" should be used to indicate the nature and theme of the literature at one time, instead of "this article" and "author" as subjects.

E standardized terms should be used instead of unknown and common symbols and terms. If there are new terms or no suitable Chinese terms, you can indicate the original text in brackets or after translation.

F. Except for being really inflexible, mathematical formulas and chemical structural formulas are generally not used, and there are no illustrations and tables.

G don't use quotations unless the document confirms or denies the published works of others.

H abbreviations, abbreviations and codes must be explained when they first appear, except those that can be clearly understood by readers of adjacent majors. Other matters that should be paid attention to when writing scientific papers, such as adopting legal units of measurement, using language and punctuation correctly, are also applicable to the preparation of abstracts. The main problems in the compilation are: incomplete elements, or lack of purpose, or lack of methods; Citation is not independent and self-evident; Improper simplification.

Step 4 introduce

The introduction of Data Analysis Report is to introduce the background and purpose of the report in a short space, put forward the realistic requirements of the study, and the preliminary work and research survey in related fields, explain the relationship between this study and the preliminary work, the current research hotspots, existing problems and the author's work significance, and lead out the theme of this paper to guide readers. A few words foreshadow the results, significance and prospect of this study, but there is no need to discuss it.

Writing points:

(1) Cut to the chase and don't beat around the bush. Avoid long stories about historical origins and achievements.