探索性数据分析（Exploratory Data Analysis）是什么？探索性数据分析的类型及示例

Explore Data Analysis (EDA) is a method that you can easily use to summarize key attributes of data. Furthermore, it is a method of extracting insights from raw data using graphics and charts. This strategy uses data visualization design and statistical data to extract insights from the data. Visualizations or charts play a key role in the business world by supporting data-backed decisions and providing insights for continuous improvement. To extract actionable insights from raw data, you need easily interpretable charts and graphics. Additionally, you need the best tools to help you access ready-made charts efficiently. Google Sheets can be considered one of the most commonly used data visualization tools as it has been around for many years and is familiar to many people. However, to perform a comprehensive Exploratory Data Analysis (EDA) using ready-made and visually appealing charts and graphs, you need to go beyond spreadsheet applications. Why? Google Sheets produces very basic charts that require additional time and effort to edit. But you don’t have to abandon Google Sheets. You can convert it into a reliable data visualization tool by installing third-party applications (add-ons). In this blog, you will learn: – How to create engaging data stories using Exploratory Data Analysis (EDA)? – Best practices related to EDA and statistics. – Additionally, you will learn about the best tools for conducting a comprehensive Exploratory Data Analysis (EDA).

What is Exploratory Data Analysis (EDA)?

According to John Tukey (the person who coined the term in the 1970s), EDA is a procedure and technique for analyzing data and interpreting results. It also involves planning, tools, and statistical data that can be used to extract insights from the raw data. You can utilize EDA to explore what data can reveal beyond hypothesis testing. Furthermore, this strategy can help you determine if the statistical techniques you are considering for analysis are appropriate.

Why is Exploratory Data Analysis (EDA) important in your business?

Until recently, understanding large and complex raw data was too daunting for us. But exploratory methods like EDA are increasingly helping us handle large datasets at scales previously unprecedented. The main advantage of EDA methods is that they allow you to investigate raw data beyond hypothesis testing. More importantly, you can use statistical models (such as mean, standard deviation, median, quartiles, etc.) to explore your data for deeper insights. Remember, there is a wealth of information hidden in data, waiting to be discovered. Even historical data collected from different sources becomes more meaningful when visualized. In today’s world, we generate a vast amount of data every day, and when used to its full potential, data can help personalize marketing communications for your business.

What are the uses of Exploratory Data Analysis (EDA)?

You can use EDA for the following operations: – Check for missing data and other errors. – Understand the data set and its underlying structure in-depth. – Validate hypotheses related to hypothesis testing. – Check for outliers, patterns, and trends in raw data. – Find parameter estimates and their related confidence intervals or margin of error.  

What are the types of Exploratory Data Analysis (EDA)?

Exploratory Data Analysis has two key variants, namely: Univariate Analysis and Multivariate Analysis. They can be either graphical or non-graphical, thus making a total of four types.

Univariate Analysis

This is the simplest form of EDA, which involves analyzing individual data points related to dimensional variables to gain insights. The main purpose of univariate analysis is to describe the data and identify patterns present. Examples of data visualization designs used in this analysis are simple bar graphs, pie charts, radial charts, etc. Visualization Source: ChartExpo

Multivariate Analysis

Multivariate analysis requires analyzing multiple variables to gain insights. The best charts used for this analysis include scatter plots, radar charts, and dual-axis line and bar charts. View examples of chart types as shown below: Scatter plot visualization: Visualization Source: ChartExpo Radar chart visualization: Visualization Source: ChartExpo

Dual Axis Line and Bar Charts:

Visualization Source: ChartExpo

How to create ready-made and insightful charts for Exploratory Data Analysis?

Google Sheets is one of the popular choices for professionals and business owners as a data visualization tool. However, its library lacks ready-made charts for EDA methods. In other words, you must invest additional time and effort to edit the charts to align with your data story. Yes, you read that right. You don’t have to waste time editing charts. You can choose to enhance your Google Sheets with third-party plugins to access ready-made and EDA-friendly charts. We recommend downloading and installing a plugin called ChartExpo in your Google Sheets. So what is ChartExpo? ChartExpo is a super intuitive plugin that you can install in your Google Sheets to access ready-made and visually appealing visualizations for your Exploratory Data Analysis (EDA). Furthermore, the EDA recommended tools also have more than 50 other ready-made advanced charts to help you succeed.

How to install ChartExpo in Google Sheets?

You can directly install the ChartExpo extension in Google Sheets from here. Once installed, you can find it in the top menu extensions of your Google Sheets application and then find ChartExpo, and click to open. Upon opening, you will see the screen below, where you can click “Create New Chart”. You will find the list of available charts through ChartExpo. You can choose any chart you want and start visualizing your data and building your own data story. In the next section, we will present examples of Exploratory Data Analysis to help you get started using an easy-to-follow method.

Exploratory Data Analysis Examples

In this section, we will introduce two main types of Exploratory Data Analysis: univariate analysis and multivariate analysis. You will also learn how to use ChartExpo to generate the most appropriate charts related to the main EDA types.

How to create different charts using ChartExpo for EDA?

Radar Chart

In this example, we will use a radar chart to visualize the table data below: | Product | Month | Orders | |—————-|———|———| | Face Cream | Jan | 80 | | Face Cream | Feb | 99 | | Face Cream | Mar | 93 | | Face Cream | Apr | 80 | | Face Cream | May | 70 | | Face Cream | Jun | 65 | | Face Cream | Jul | 85 | | Face Cream | Aug | 90 | | Face Cream | Sep | 80 | | Face Cream | Oct | 75 | | Face Cream | Nov | 65 | | Face Cream | Dec | 80 | | Whitening Cream| Jan | 100 | | Whitening Cream| Feb | 60 | | Whitening Cream| Mar | 95 | | Whitening Cream| Apr | 75 | | Whitening Cream| May | 100 | | Whitening Cream| Jun | 60 | | Whitening Cream| Jul | 95 | | Whitening Cream| Aug | 75 | | Whitening Cream| Sep | 109 | | Whitening Cream| Oct | 80 | | Whitening Cream| Nov | 109 | | Whitening Cream| Dec | 75 | | Beauty Cream | Jan | 50 | | Beauty Cream | Feb | 55 | | Beauty Cream | Mar | 51 | | Beauty Cream | Apr | 40 | | Beauty Cream | May | 45 | | Beauty Cream | Jun | 30 | | Beauty Cream | Jul | 39 | | Beauty Cream | Aug | 45 | | Beauty Cream | Sep | 56 | | Beauty Cream | Oct | 39 | | Beauty Cream | Nov | 48 | | Beauty Cream | Dec | 44 | Copy and paste the data into Google Sheets to start creating visualizations for Exploratory Data Analysis. Search for “Radar Chart” on the toolbar. Select the sheet containing the data. Fill in your metrics and dimensions. In our example, the key metric to fill in is the Orders. For dimensions, fill in the variables: Product and Month. Click on the “Create Chart” button to complete the data visualization using a radar chart. Insights The best-performing product is Whitening Cream, as it outperforms Face Cream and Beauty Cream in the top months. The worst-performing product is Beauty Cream. Face Cream performs better than Whitening Cream in months of January, March, May, July, and November.

Pareto Chart

In this example, we will use a Pareto chart to visualize the table data below: | Product | Sales Volume | |————–|————-| | Blush | 1579 | | Mascara | 1962 | | Lipstick | 3654 | | Foundation | 2578 | | Powder | 4942 | | Eyebrow Pencil| 5561 | | Eyeshadow | 2961 | | Nail Polish | 4831 | | Lip gloss | 8961 | Copy the data (table above) and paste it into Google Sheets to create a Pareto chart. Search for “Pareto Chart” on the toolbar. Fill in your metrics and dimensions. In our example, fill in the key metric: Sales Volume. For dimensions, fill in the variable: Product. Click on the “Create Chart” button to complete the visualization process. Insights Lip gloss, Eyebrow Pencil, Powder, Nail Polish, and Lipstick are the 20% products that drive 80% of total sales for the brand. Lip gloss alone accounts for 24% of the cumulative sales volume. Eyebrow Pencil accounts for 39% of the total sales volume. In this section, we will use a dual-axis line and bar chart (one of the exploratory data analysis-friendly visualizations) to analyze the dataset below. Let’s dive in. | Quarter | Sales Volume | Growth | |———–|————–|——–| | Q1-19 | 7000 | 4.2 | | Q2-19 | 7606 | 7.6 | | Q3-19 | 7895 | 3.8 | | Q4-19 | 8242 | 4.4 | | Q1-20 | 8327 | 0.7 | | Q2-20 | 8768 | 5.3 | | Q3-20 | 9337 | 6.5 | | Q4-20 | 9589 | 2.7 | Export the above table data to Google Sheets for a dual-axis line and bar chart. Search for “Dual Axis Line and Bar Chart” in the toolbar. Fill in your metrics and dimensions. In our example, the key metrics to fill in are Sales Volume and Growth. For dimensions, fill in the variables: Click the “Create Chart” button to complete the simple process. Insights The best-performing quarter is Q2-19 as the growth exceeds the sales volume for the period. On the other hand, the worst-performing period is Q1-20.

Advantages of Exploratory Data Analysis

EDA methods are flexible and can adapt to changes as data analysis progresses. Additionally, it can provide a solid foundation for your analysis and storytelling tasks.

Applications of Exploratory Data Analysis (EDA)

You can use exploratory methods to measure central tendencies, which can provide you with an overview of both univariate and multivariate variables. Central tendencies are measures of averages, medians, modes, and ranges.

FAQs:

What is Exploratory Data Analysis?

Exploratory Data Analysis is a statistical method used to investigate patterns, trends, and anomalies in raw data. It involves planning, tools, and statistical data that can be used to extract insights from the raw data. You can use EDA to explore what data can reveal beyond hypothesis testing.

What is the use of Exploratory Data Analysis?

Exploratory Data Analysis (EDA) is a method of analyzing a data set to summarize its main characteristics using visualization designs such as tables, charts, and graphs. You can use this method to measure central tendencies (mean, median, mode, range).

What is the purpose of EDA?

The main goal of EDA is to help you uncover hidden insights in your data set. You can use this method to check for missing data and other errors. Additionally, you can use this method to gain an in-depth understanding of your data set and its underlying structure.

What are the two main objectives of Exploratory Data Analysis (EDA)?

Exploratory Data Analysis (EDA) has two main objectives, namely: – Checking for missing variables and other errors that may distort key insights. – Obtaining hidden insights such as trends, outliers, and patterns in the raw data. Additionally, EDA is flexible and can adapt to changes as needed. Pack it up Exploratory Data Analysis (EDA) is a statistical-based method used to analyze data and interpret results. It also involves planning, tools, and statistical data that can be