Creating an Intelligent Analysis Assistant

Scenario

The intelligent analysis assistant provides conversational data analysis, allowing users to interact with data through natural language to perform tasks such as data analysis, gaining intelligent insights, and creating visual charts. The intelligent analysis assistant not only simplifies data analysis but also enhances the efficiency of self-service chart analysis. In this section, we will use sales data of a specific product as an example to demonstrate the method of using the intelligent analysis assistant.

Procedure

Figure 1 Process of creating an intelligent analysis assistant

Data Description

To protect user privacy and data security, all data has been sampled and anonymized. In this example, there are three data tables: order_info_demo_01 (order table), product_info_demo_02 (product information table), and user_info_demo_03 (user data table). For details, see Table 2.

Step 1: Create a Project

1. Log in to the DataArts Insight console.
2. Click in the upper left corner of the management console to select a region and select an enterprise project from Enterprise Project in the lower left corner of the Workspace page.
3. Click Create Project in the right pane of the console.
4. Enter the project name, visible type, members, and description. See Figure 2.
5. Set the parameters and click OK.
   Figure 2 Creating a project
   

Step 2: Connect to a Data Source

In this example, a GaussDB(DWS) data source is connected through the public network..

1. Select the desired project and click the name to access the project space.
2. Choose Data Management > Data Sources. On the displayed page, click Create Data Source. In the slide-out panel, set Access Network Type to Public network. See Figure 3.
3. Enter the name, domain name, port number, and database, and click Test Connection. For details about the parameters, see Table 3. Once the connection test is successful, the data source is successfully connected.
   Figure 3 Clicking Create Data Source
   

   Table 3 Parameters

   Parameter	Mandatory	Description
   Source Database Type	Yes	Type of the accessed data source. In this example, GaussDB(DWS) is selected.
   Access Network Type	Yes	In this example, Public network is selected.
   Name	Yes	Display name of the data source in the list.
   Domain Name	Yes	IP address of the data source.
   Port	Yes	Port number for logging in to the database.
   Database	Yes	Name of the database to be logged in to.
   SASL_SSL	-	It is used for trusted identity authentication and secure data transmission when DataArts Insight retrieves data from the data source. This function is enabled by default.
   Username	Yes	Username for logging in to the database.
   Password	Yes	Password for logging in to the database.

Step 3: Create a Dataset

1. Select the desired project and click the name to access the project space.
2. Click next to the data source name and select Uploaded Files (Figure 4). Select the Excel file you want to upload, modify parameters as required, and click OK. Repeat this step to upload all data tables in sequence. In this example, there are three data tables. For details about the data tables, see Data Description.
   Figure 4 Uploading a file
   
   

   Before uploading files, you need to configure the data storage bucket on the System Settings > OBS Bucket Settings page.

3. Choose Data Management > Datasets. On the displayed page, click Create Dataset.
4. Under My Data Sources, select the data source that is connected to in step 2. Click the Uploaded Files and double-click or drag a data table to add it to the operation panel.
   Figure 5 Selecting a data table
   

5. Double-click the data table name and set the table's association in the operation panel.
   • Set up a left outer join between product_info_demo_02 and order_info_demo_01 using the id field as the key for the association.
   • Set up a left outer join between product_info_demo_02 and user_info_demo_03 using the id field as the key for the association.
   Figure 6 Creating a model by associating multiple tables
   
6. Click the Visual Configuration tab and enter descriptions of the fields.
   Figure 7 Visual configuration
   
7. Click the Data Preview tab and then Refresh Preview to view basic information about the dataset.
   Figure 8 Data Preview tab
   

8. Confirm the information and click Save.

Step 4: Create an Intelligent Analysis Assistant

1. Click in the upper left corner to select a region.
2. In the lower left corner of the navigation pane, select an enterprise project from Enterprise Project.
3. Click the name of a project in the My Projects list on the right.
4. In the navigation pane on the left, choose Data Analytics > Intelligent Analysis Assistants.
5. Click Create Domain. In the dialog box that appears (Figure 9), enter a name and select an associated dataset.
   Figure 9 Creating an intelligent analysis assistant
   

6. Click OK.
7. In the list, locate the intelligent analysis assistant you created, and click Train in the Operation column.
   If Training Status is Successful, as shown in Figure 10, the intelligent analysis assistant model is successfully trained.

   Figure 10 Training the intelligent analysis assistant
   

Step 6: Use the Intelligent Analysis Assistant

1. Click the name of the desired project in the My Projects area in the right pane.
2. In the navigation pane on the left, choose Data Analytics > Intelligent Analysis Assistants.
3. Locate the intelligent analysis assistant you trained and click Q&A in the Operation column.
   Figure 11 Intelligent analysis assistant's Q&A page
   

4. Input the question "What are the prices of different types of products?" on the Q&A page.
   Figure 12 What are the prices of different types of products