The Intelligent Data Insight Analysis Assistant (hereinafter referred to as the DataWind Analysis Assistant, Analysis Assistant, or AI Assistant) is a self-service analysis intelligent assistant. During the process of self-service data retrieval and chart creation, users may need to create a large number of charts for various business scenarios. The Analysis Assistant not only helps users quickly generate and modify field expressions required for chart creation, but also assists users in analyzing the query results of the current chart through secondary analysis. By simply describing requirements in natural language, the Analysis Assistant can help users quickly complete data retrieval and chart building tasks.

Secondary analysis: Click the analysis configuration area, then select Secondary Analysis. If a dialog box appears as shown below, it indicates that secondary analysis has been successfully enabled.

3.1 Field expression generation #

3.1.1 Generate expressions #

When adding a new field in a visual query, for the step of defining the field expression, the Analysis Assistant can provide suggestions for expressions based on your natural language description and help you correct incorrect expressions.
Step 1: In the visual query editing interface, at the upper left of the field list area, click the operator to the right of the dataset name.

Step 2: In the field expression box, describe your requirements for generating the expression.

3.1.2 Expression explanation #

You can move the mouse over the generated expression, and the AI Assistant will provide the logic for generating the expression. You can learn how to write the corresponding type of expression by reviewing the logic.

3.1.3 Example scenarios #

Scenario 1: Calculate the total sale amount of Shanghai

Scenario 2: Calculate the time progress of the current quarter
In addition to generating expressions based on fields in the dataset, the AI assistant can also generate some general field expressions or field expressions that do not involve dataset content.
For example, to calculate the time progress of a quarter, it is necessary to use date functions related to quarters. Searching for how to use such functions is often time-consuming. In this case, users can describe their requirements to the AI assistant, saving the time needed to search for function details.
(1) Calculate the number of days that have passed in the current quarter

(2) Calculate the total number of days in the current quarter

3.2 Chart generation #

This feature supports generating charts based on your natural language description. When generating charts, you can use analysis features and create expressions.

Secondary analysis allows you to analyze the query results of the current chart using natural language and supports saving the results to the dashboard.

Click Analyze and select Secondary analysis to ask questions in the dialog box.

4.1 View analysis history #

Secondary analysis supports multi-turn conversations. You can use the "History analysis steps" button on the right side of Secondary analysis to view all steps of the current round, and you can also click the "Apply" button after a step to restore the chart to its state at that step.

4.2 Secondary calculation and analysis of derived fields from analysis features #

After applying the analysis features of visual queries, additional metric fields, that is, derived fields, will be generated, such as period-over-period growth rate, period-over-period raw value, metric percentage, and so on. Secondary analysis enables secondary calculations on derived fields.

4.2.1 Calculations between derived fields #

(1) Add a new column for the combined period-over-period growth value of income and cost, calculated by adding the period-over-period growth values of income and cost.

(2) Add a new column for the ratio of the percentage of sales amount to the percentage of profit, with values rounded to two decimal places.

4.2.2 Calculations between derived fields and dataset fields #

Add a new column for the growth coefficient, which calculates the ratio of the month-over-month sales growth value to the sales.

4.2.3 Derived field filtering #

Filter data where the month-over-month sales growth rate is greater than 5%.

4.3 Multi-round aggregation operations #

Secondary analysis can achieve multi-round aggregation based on different dimensions, such as requirements for stratification by a specific metric.
(1) Add a new column: provinces with income greater than 900,000 are classified as high-income provinces, and those with less income are classified as low-income provinces:

(2) Calculate the number of provinces in each income level and their total income, then remove the province column:

4.4 Enhanced filtering capabilities #

After applying the table calculation feature, secondary analysis can filter the final results to further analyze specific dimensions.

4.4.1 Dimension filtering based on the final chart results #

View only the data for the Northeast region:

4.4.2 Filter data that meets multiple field conditions simultaneously #

Filter for provinces that rank in the top three for both income and order quantity:

4.4.3 Simultaneously filter two types of fields #

Filter provinces that are both Zhejiang and sell "Road 15" type of products:

4.5 Add virtual columns to the pivot table #

Secondary analysis enables calculations between different enumerated values under the same column dimension in the pivot table.
For example: add a new column named Cost, which calculates Sales revenue minus Profit.

4.6 Enhanced table calculation capabilities #

Secondary analysis supports more calculation methods for table calculations and also supports multi-metric calculations.

4.6.1 Calculate the ratio between column fields #

Add a new column for the ratio to the most recent day's profit, calculating the ratio of profit to the most recent day's profit, rounded to two decimal places.

4.6.2 Cross-row calculations based on multiple metrics #

Add a new column named Difference between today's income and the previous day's cost, which subtracts the previous day's cost from the current date's income, rounded to two decimal places.

4.6.3 Aggregate long-tail metrics into a single row for display #

Combine provinces with income less than 1,000,000 into a single row named Other, sum the income column, and for the income summary column in this row, use the sum of the income column for all rows in the current result.

4.7 Perform merged row operations based on identical dimension enumeration values #

There is an event dataset containing event A and event B. Event A contains id and timestamp a, and event B contains id and timestamp b.
Metric to calculate: For the same id in events A and B, timestamp b minus timestamp a. To implement this requirement, use the id to match the two events and subtract the reported timestamps of the two events.
For example: merge rows with the same id, keep one row in the association result.

4.8 Custom rule sorting #

When there are many dimension enumeration values in the chart, secondary analysis allows you to quickly sort the chart according to the specified order of enumeration values.
For example: place Liaoning at the top among provinces, and sort the rest in descending order of sales amount.

(1) All content output by this service is generated by artificial intelligence models.
(2) Applicable users of this feature: limited to customers for internal, proprietary business, and personal use only. Customers are not allowed to provide generative artificial intelligence services to the public through the Analysis Assistant.
(3) This feature is a value-added module and supports both SaaS and on-premises deployment versions of the product. To purchase or try it, please contact the sales staff for consultation.
(4) Currently, the AI capabilities of DataWind Intelligent Data Insight support the following types of datasets:

• Supports extracted datasets and supports datasets uploaded as Excel/CSV files.
• Supports direct connection datasets, currently including only: ClickHouse / ByteHouse, Doris / selectdb.