Don't have an analytical idea in the face of a problem? Detailed Analytical Thinking for Data Analysts (Part I)

In today's business environment, data has become one of a business's most valuable assets. With the rapid development of big data, artificial intelligence, and machine learning, we are able to extract valuable information from massive amounts of data that is critical to guiding business decisions, optimizing business processes, improving customer experiences, and driving innovation. However, turning data into actionable insights relies on effective data analytics methods.

Data analytics is not only a technical tool, but also a way of thinking that helps us understand the patterns, trends, and associations behind data. Through data analytics, companies can anticipate market changes, evaluate the effectiveness of marketing campaigns, optimize product features, and develop more accurate business strategies. In this process, it is particularly important to master the commonly used data analysis methods.

This article will delve into a range of commonly used data analysis methods to provide you with a comprehensive analysis toolbox. We'll cover the principles of each approach and how to implement it to help you develop your data analytics skills to succeed in a data-driven business world.

Whether you're a data analyst, market researcher, product manager, or business decision-maker, this article will provide you with valuable knowledge and practical guidance. Let's embark on a journey of data analytics and discover how to harness the power of data to drive growth and innovation.

1. Why do we need data analysis methods?

Problem 1: Lack of awareness of data-driven decision-making

Decisions at work often rely on intuition rather than data-based analysis. This practice can lead to the following consequences:

• Created a large number of articles, but were unable to identify which topics or types of articles would best engage and meet user needs;
• Invested in multiple paid promotion channels, but was unable to measure which channels generated the greatest market response and returns;
• Multiple product features were developed, but it was difficult to determine which ones really enhanced the user experience and product value.

They rely on subjective judgment rather than a decision-making process supported by objective data, which may be one of the reasons why they have not been able to advance in their careers for a long time.

Question 2: Statistical data analysis

While numerous charts and reports were produced, they failed to dig deep and solve the core problems in the business. Graphs are made using data, but analysis tends to be superficial, summarizing and presenting what is already known.

For example, a data analyst may report that "sales are down this month compared to the previous month", but fail to further analyze the specific factors that contributed to the decline in sales, such as market competition, product issues, and insufficient marketing strategies, so they can't come up with a practical solution. When asked more in-depth questions, such as "What is the reason for the decline in sales?" Or, "How should we adjust our strategy to increase sales?" They are often unable to give satisfactory answers.

Problem 3: Relying too much on data analysis tools and lacking actual analysis capabilities

Although they have mastered a variety of data analysis tools, such as Excel, SQL or Python, and are familiar with how to use these tools, it is difficult to effectively apply these tools for in-depth analysis when faced with specific business problems.

Such people may focus too much on the skill of the tool and lose sight of the real purpose of data analysis – solving real problems. They may be proficient at using tools to generate charts and reports, but they don't perform as well as they could when it comes time to extract valuable insights from data, identify the source of problems, or predict future trends.

The importance of data analysis methods

When faced with problems, we often have all kinds of scattered ideas that lack organization and direction. If you can systematize these disorganized ideas and form a clear line of thought, then the efficiency of solving problems will be greatly improved. This requires the application of analytical methods. Mastering these methods is equivalent to having the ability to integrate fragmented information and solve problems effectively.

It is within the framework of an analytical approach that we understand how to properly use technical tools to analyze data to solve specific business problems. Without the right analytical methods as a guide, the use of technical tools can become blind and ineffective and fail to achieve the intended purpose of analysis. Therefore, the analytical approach not only provides us with a framework for solving problems, but also a prerequisite for the effective use of technological tools.

2. What are the data analysis methods?

1. 5W2H analysis method

5W analysis

The 5W analysis is a systematic approach to problem-solving, which requires us to ask five basic questions about any phenomenon: first, ask "what" to clarify the essence of the problem; This is followed by "When" to determine when the problem occurred; Then there's "Where" to identify where the problem is occurring; This is followed by "Why" to explore the cause of the problem; Finally, there is the "Who" to understand the person in relation to the issue.

2H Supplementary Questions

On the basis of 5W, 2H further deepens the analysis of the problem, which includes two key questions: "How" focuses on the method and process of solving the problem; "How Much" focuses on the cost or resources needed to solve the problem.

Application scenarios

• Project management: During the project planning phase, 5W2H can help the project team clarify the project goals, timeline, location, people involved, why, methodology, and budget.
• Market research: Through 5W2H analysis, you can have a comprehensive understanding of consumer behavior, market trends, competitor strategies, etc., so as to guide the formulation of market strategies.
• Product development: During the design and development of new products, 5W2H helps to identify product requirements, target users, usage scenarios, design reasons, functional features, and cost budgets.
• Troubleshooting: In the technical or production world, 5W2H is a common tool for troubleshooting and root cause analysis, helping to quickly locate problems and find solutions.

Although 5W2H analysis is easy to understand and apply, it may not be sufficient to provide a comprehensive solution when dealing with complex business problems. This is because complex business problems are often not caused by a single factor, but are the result of a combination of factors.

Taking "why is sales declining" as an example, the decrease in sales may be the result of a combination of factors, such as market competition, product characteristics, marketing strategies, economic environment, changes in consumer preferences, etc. In this case, a single 5W2H analysis may not be able to dig deep into all relevant factors, so other, more specialized analysis methods are needed.

The data analysis template mentioned in the example is shared with you——

https://s.fanruan.com/x3k5k

Zero-based quick start, but also according to the needs of personalized modifications

2. Industry analysis methods

Industry analysis is a critical process that can be applied to a variety of scenarios:

• Personal career planning: when individuals are considering career development paths, assessing the potential of different industries and the fit of personal career goals;
• Corporate strategic planning: when an enterprise needs to have an in-depth understanding of the external environment, including market trends, industry dynamics, and competitors, in order to formulate or adjust its long-term development strategy;
• Solving big problems: When faced with a major challenge or opportunity in the industry, an in-depth analysis of the industry structure, drivers, and potential issues is required.

One of the effective ways to conduct industry analysis is to adopt the PEST analysis framework. PESTLE analysis is a macro environment analysis tool that is widely used in industry analysis. It focuses on evaluating the external factors influencing the company's development from the following six dimensions:

Political: Analyze the impact of government policies, laws and regulations, political stability, and the political environment on the industry.

Economic: Examines how macroeconomic indicators, economic growth, exchange rates, inflation rates, and other factors affect the industry.

Social: Assess the potential impact of factors such as demographic changes, cultural trends, consumer attitudes, and lifestyles on the industry.

Technological: Explore how technological advancements, innovations, and disruptive technologies are shaping the future of the industry.

Legal: Deals with laws and regulations, compliance requirements, intellectual property protection, contract laws, and legal changes that may affect your business.

Environmental: This includes natural resources, ecosystems, climate change, environmental policy and sustainability issues, as well as their impact on business operations.

Through PESTLE analysis, we can fully understand the macro environment of the industry, identify opportunities and threats, and provide a solid basis for strategic decision-making.

(1) How to use PESTLE to carry out industry analysis

政治分析（Political）

Policy environment analysis is an important part of evaluating the impact of government policies and laws on a company's operations. This analysis can be conducted with a deep dive around the following core questions:

Legal and regulatory review: First, identify all laws and regulations that are directly related to the company's business. This includes, but is not limited to, industry-specific regulations, labor laws, environmental laws, etc.

Impact assessment: For each law, analyze its specific impact on the company. This can involve aspects such as compliance costs, operational constraints, potential legal liabilities, and more.

Investment Policy Analysis: Study the investment incentives, subsidies, and regulations provided by the government that may affect the company's investment decisions.

Tax Policy Considerations: Understand the latest tax regulations, including tax rate changes, tax incentives, exemptions, etc., and assess the potential impact of these changes on your company's financial position.

经济分析（Economic）

Economic analysis should be considered from the following aspects: macroeconomics: examine macroeconomic indicators such as economic growth, inflation rate, exchange rate, and interest rate; Consumer confidence: Understand the impact of consumer confidence and purchasing power on market demand. Economic Cycle: Analyze the impact of different stages of the economic cycle on the industry.

Social Analysis (Social)

The following factors are taken into account in conducting social analysis: Demographics: the study of demographic changes, such as age distribution, sex ratio, education level, etc. Cultural Trends: Assess the impact of changes in culture, values, and social attitudes on consumer behavior. Lifestyle: Learn about lifestyle changes such as health awareness, work-life balance, etc.

技术分析（Technological）

The technological environment refers to the impact of external technology on the development of a company. The factors that mainly consider these aspects: Technological progress: pay attention to the technological development in the industry, such as emerging technologies, product innovation, etc. Automation and Artificial Intelligence: Assess the impact of automation and artificial intelligence on industry productivity and employment. Information Technology: Analyze how information technology is changing industry operating models and consumer behavior.

Legal Factor Analysis (Legal)

Laws and regulations: Study the legal environment that affects the industry, such as antitrust law, consumer protection law, etc. Compliance requirements: Understand the compliance standards and potential legal risks that businesses must adhere to.

环境因素分析 (Environmental)

Environmental policy: Assess the impact of environmental regulations on the industry, such as emission standards, resource use restrictions, etc. Sustainability: Understand how sustainability trends are impacting your business's operations and product development.

3. Comparative analysis methods

A contrastive analysis method is an analytical technique that identifies differences, trends, and patterns by comparing data from different objects or at different points in time. This method is widely used in business, scientific research, social sciences, and many other fields.

When conducting a comparative analysis, two core issues need to be clarified: first, determine the object of comparison; Second, decide on the method of comparison.

(1) Selection of comparison objects

There are usually two options for comparison: to compare with their own historical data, or to compare with industry standards.

• Compare with your own historical data

When comparing and analyzing historical data, companies can evaluate their own growth, efficiency, and market performance by comparing performance metrics over different time periods.

• Comparison with industry standards

When faced with a problem, distinguishing whether it is a general trend in the industry or a company-specific situation can be achieved by comparing it to the industry average.

(2) How to compare

We've explored the first key issue of the contrastive analysis method – choosing who to compare. Now, let's dive into the second question – how to compare. In general, comparisons can be made in the following three dimensions:

• Comparison of data size

The first step in the comparison is to assess the overall size of the data. This can be achieved by calculating the mean, median, or specific business metrics to get a representative statistic of the dataset.

• Analysis of data volatility

To measure the volatility of data, a commonly used statistical tool is the coefficient of variation, which is obtained by dividing the standard deviation by the mean to reflect the relative volatility of the data.

• Identification of trend changes

The analysis of trend changes involves observing the evolution of data over time. This can be achieved in several common ways:

Time series analysis: By drawing a time line chart, we can use time as the horizontal axis and data volume as the vertical axis, so as to visually observe the changes of data over time. Time line charts not only help us understand historical trends, but also serve as a basis for predicting future trends.

Month-on-month analysis: Month-on-month comparison refers to the comparison with the data of the previous period (such as last week, last month), and it is suitable for analyzing data fluctuations and cyclical changes in the short term. For example, by calculating the percentage decrease in December 2020 compared to November 2020, we can get the month-on-month rate of change.

Year-on-year analysis: Year-on-year comparison, which involves comparing data to the same period in the previous year (e.g., the same month of the previous year), is more suitable for analyzing the impact of long-term trends and seasonality on the data.

Through these methods, comparative analysis can help us deeply understand the essence of data, reveal the differences and connections between different time periods or different objects, and provide powerful data support for decision-making.

(3) Precautions

To ensure comparability, the objects or time periods being compared should be comparable. When making comparisons, it is important to note that the size of the objects to be compared is consistent. For example, when comparing the financial health of two companies, you need to make sure that they are in the same industry, similar in size, and that the financial metrics compared are calculated in the same way.

4. Hypothesis testing and analysis methods

The core idea of the hypothesis testing analysis method can be summarized as a decision-making process based on logical reasoning, which is similar to the principle of presumption of innocence in legal proceedings. In court, the judge usually presumes the innocence of the defendant when hearing a case. Subsequently, it is the responsibility of the prosecution to gather and present evidence to convince the judge or jury to accept the defendant's guilt. (The hypothesis testing analysis method mentioned here is not a construction test method in a strict statistical sense.)

(1) Steps

Applying this logic to data analysis, the hypothesis testing analysis method involves the following three main steps:

• Formulate hypotheses

First, an initial hypothesis about the research question is formed, which usually consists of a null hypothesis and an alternative hypothesis. The null hypothesis usually indicates that the observed phenomenon can be attributed to a chance factor, while the alternative hypothesis indicates that there is some kind of effect or difference.

• Collect data

After the hypothesis is clarified, data collection is carried out. These data will serve as a basis for testing hypotheses. What data you're looking for is related to the hypothesis you're trying to validate

• Draw conclusions

The collected data is analysed using statistical tools and methods to determine whether the hypothesis is valid. The results of the analysis will guide the final decision-making process to determine whether to accept the null hypothesis or the alternative hypothesis.

Hypothesis testing requires us to explicitly formulate hypotheses and then test the validity of those hypotheses through logical reasoning. This process forces us to think in a systematic and structured way, thus exercising and improving our logical thinking skills. Through hypothesis testing, we learn how to make more rigorous decisions based on data and statistical principles, rather than just intuition or bias.

Hypothesis testing analysis methods allow us to identify and evaluate the influence of different factors on observed phenomena or outcomes, which is especially important in the exploration of causality. When faced with a complex problem, hypothesis testing can help us gradually narrow down the potential causes by ruling out impossible explanations.

(2) How to objectively formulate hypotheses

From the three key dimensions of users, products, and competing products

In order to fully check whether the hypothesis is comprehensive, we can start from three key dimensions: users, products, and competitors, which correspond to different departments within the company: the user dimension is associated with the operation department, the product dimension corresponds to the product department, and the competitor dimension is closely related to the marketing department. Based on these three dimensions, we can construct the following hypotheses:

• User problem hypothesis: If you suspect that the problem originates from the user level, you can deeply analyze the source channel of the user, or identify possible problem points by drawing the operation flow chart of the user's use of the product.
• Product Problem Hypothesis: If the problem may be related to the product, research should be done to see if the product currently being sold actually meets the market demand and user expectations.
• Competitor Impact Hypothesis: If the problem is caused by the competitive environment, you need to pay attention to whether the competitor has launched a promotional campaign to attract users, resulting in customer churn.

Starting from the 8P marketing theory

此外，我们还可以借助经典的8P营销理论（Product, Price, Place, Promotion）来提出假设，涉及营销组合的八个基本要素：

• Product: It involves the tangible or intangible products provided by the company to the target market, including the product itself, brand, packaging, style, service, technology and other aspects.
• Price: The price strategy of the user when purchasing the product, covering basic pricing, discounts, payment terms, and various pricing methods and techniques.
• Place: Describes the sales channel and path of a product from the producer to the consumer.
• Promotion: A variety of promotion methods adopted by enterprises to stimulate consumers' desire to buy and promote sales growth, including advertising, sales team promotion, sales promotion activities, etc.
• People: Emphasize the role of employees in service marketing, including their service attitude, skills, and interactions with customers.
• Process: involves the entire process of customer experience, including the design and management of service processes, to ensure the efficiency of the process and customer satisfaction.
• Physical Evidence: In service marketing, it refers to the service environment and all the tangible elements that can affect the customer's perception of the quality of service.
• Performance: This P is sometimes used to replace or add to a model to emphasize the efficiency and effectiveness of the product or service provided, and how to meet or exceed customer expectations.

In order to identify the reasons for the decline in sales performance, we can put forward hypotheses from eight different perspectives: product, price, channel, promotion, personnel, process, physical evidence, and performance according to the 4P marketing theory, and investigate and analyze them one by one. This approach helps us to take a holistic view of the market situation, identify the root cause of the problem, and develop effective strategies to deal with it.

Formulate assumptions from the business process

(3) Precautions

When applying hypothesis testing analysis methods, the following key points should be taken into account to ensure the rigor and validity of the analysis:

• Evidence-based conclusions: In reaching conclusions in the third step of hypothesis testing, we should not rely solely on subjective guesses, but on the evidence gathered to support or refute our hypothesis.
• Iterative analysis process: Hypothesis testing is not a one-time activity, but a process that requires continuous iteration. Even after the initial conclusions have been reached, the analytical work should not stop. We need to ask questions and validate them with data, repeating what-if analyses until we get to the root cause of the problem.
• Use a combination of multiple analytical methods: In the process of hypothesis testing, we should not limit ourselves to a single analytical approach, but should use a combination of other analytical tools and techniques to obtain more comprehensive and in-depth insights.
• Build an analytical framework: Before you begin your analysis, it's a good idea to draw a hypothesis testing analysis diagram that connects the questions, hypotheses, and required data in a logical order to show the full picture of the analysis from top to bottom. Doing so helps to clear your thinking and ensures that each step of the analysis process is organized and well-founded.

By focusing on these points, we can apply hypothesis testing analysis methods more systematically and scientifically, improve the accuracy and reliability of analysis, and thus provide solid data support for decision-making.

5. Relevant analysis methods

Correlation analysis is a statistical method used to explore whether there is a relationship between two or more types of data, as well as the strength and direction of that relationship. In the field of data analysis, the problems we face can be divided into two broad categories depending on the type of data required:

• Single-type data-type research: Some questions involve only a single type of data, which is analyzed independently. For example, when looking at a specific characteristic of a person's height, we focus on the height data itself.
• Multi-type data-based studies: Other studies need to explore the relationship between two or more different types of data. For example, when analyzing whether there is a link between height and weight, we need to use correlation analysis.

Correlation analysis is a statistical method used to assess whether there is a certain relationship between two or more variables, as well as the strength and direction of that relationship.

• Correlation: If the analysis shows some degree of correlation between the two types of data, we call it "correlated". This may mean that as one variable increases, another also tends to increase (positive correlation), or that an increase in one variable is accompanied by a decrease in another variable (negative correlation).
• No correlation: Conversely, if the data shows that there is no obvious connection between the two variables, i.e., changes in one variable have no effect on the other, we call it "no correlation".

Through correlation analysis, researchers can identify potential connections between variables, which can provide a basis for further in-depth research and decision-making. It is important to note that correlation does not imply causation, it only indicates that there is some statistical connection between variables.

(1) What are the uses of relevant analysis?

• Relationship identification: When investigating whether there is a connection between multiple data sets, correlation analysis can help us identify potential connections between these data.
• Broadening perspectives: In the process of problem solving, it expands our analytical horizons and prompts us to consider the interactions between multiple types of data beyond a single piece of data.
• Communication & Understanding: The concept of correlation analysis is easy to understand and easy to communicate with non-specialists, helping to gain understanding and support from team members and decision-makers.
• Comprehensive applications: It can be used in conjunction with other analysis methods, such as regression analysis, for deeper data exploration and pattern recognition.

(2) Correlation coefficients are used to measure correlations

In statistics, a correlation coefficient is a numerical metric used to quantify the strength and direction of a linear relationship between two variables, usually denoted by the letter r. It plays an important role in analyzing data relationships and has two core functions:

Quantitative correlation: The numerical magnitude of the correlation coefficient accurately reflects the degree of correlation between two datasets.

Indicates the direction of correlation: The positive and negative signs of the correlation coefficient reveal the trend relationship between the variables, i.e., whether they change in the same direction or in the opposite direction.

The correlation coefficient can range from -1 to 1, where -1, 0, and 1 represent the three extremes of the correlation coefficient, which are explained as follows:

• When the correlation coefficient r = 1, it means that there is a completely positive linear relationship between the two variables, i.e., an increase in one variable is accompanied by an increase in the other.
• When the correlation coefficient r = -1, it means that there is a completely negative linear relationship between the two variables, i.e., an increase in one variable leads to a decrease in the other.
• When the correlation coefficient r = 0, it is traditionally considered that there is no linear relationship between two variables, but this does not rule out that there may be other types of nonlinear relationships between them.

In a scatter plot, we can intuitively understand the meaning of the correlation coefficient by the distribution pattern of the data points. The greater the absolute value of the correlation coefficient, the stronger the linear relationship between the two variables.

In order to determine whether there is a "correlation" between two variables in practical application, the magnitude of the correlation coefficient is usually classified according to the following criteria:

• Low correlation: When the absolute value of the correlation coefficient is between 0 and 0.3.
• Moderate correlation: When the absolute value of the correlation coefficient is between 0.3 and 0.6.
• High correlation: When the absolute value of the correlation coefficient is between 0.6 and 1.

These classifications provide a general way to assess correlations between variables. However, it is important to note that correlation does not imply causation. Even if two variables are highly correlated, it cannot simply be concluded that one variable is the cause of a change in the other. In some cases, more complex methods, such as multiple regression analysis, may be required to control for the effects of other variables and to more accurately identify causal associations between variables.

When performing correlation analysis, scatter plots are a common visualization tool that visualizes the correlation between two variables. In addition, correlation analysis can help us identify which factors have significant relevance to the research objectives.

(3) Precautions for related analysis

• Correlation vs. causality: It's important to note that correlation is not the same as causation. Just because there is a correlation between two variables, it doesn't mean that one variable is the cause of the change in the other.
• Univariate control method: In order to determine whether there is a causal relationship between two variables, you can use the "univariate control method", that is, change one factor while controlling all other factors and observe the effect of this change on the outcome.
• Use case: In most cases, while we may not be able to directly determine cause and effect, we can also provide valuable information by identifying correlations. However, for complex issues that require a deep understanding of the reasons behind the event, we may need to further explore and verify the underlying causal relationships on the basis of discovering correlations.

III. Summary

In the digital age, the value of data has transcended its original form and has become the core of business decision-making. Through the exploration of this article, we understand that data analysis is not only a technology, but also a way of thinking that deeply understands the business and reveals the truth of the market. We've discussed a range of data analytics methods in depth, each of which is a powerful tool to help us extract insights from our data.

As technology continues to advance, so does the approach to data analysis. As users and interpreters of data, we need to keep learning and adapting to new tools and technologies to keep our analytical skills up to date. Remember, the purpose of data analytics is always to better understand the world and use that understanding to create better products, better services, and more informed decisions.

With its superior performance and flexibility, FineBI data analysis tools provide enterprises with a comprehensive data insight platform. It supports the rapid processing and analysis of massive amounts of data, ensuring immediate business decision support. Users can easily create complex data models through FineBI's drag-and-drop interface, and its rich visualization components make the presentation of data vivid and intuitive. In addition, FineBI's real-time data processing capabilities enable enterprises to keep up with the pulse of the market and respond quickly to changes. Whether it's a data analyst, business manager, or decision-maker, FineBI provides the deep analytics and instant insights they need to stay ahead of the data-driven business competition.