Data preparation, which includes data connection and modeling, forms the foundation of data visualization analysis. Typically, users need to perform data connection before conducting visualization analysis.
Data connection is the first step in interfacing with a database. After this step is completed, you can create datasets to serve as input for data visualization query analysis, or as input for data visualization modeling. Visualization modeling refers to basic data processing, data cleansing, modeling, and other operations performed on existing data. The datasets processed in this way can be used for data visualization query analysis. IT personnel, data developers, or data analysts typically perform more operations during the visualization modeling phase.
The logic among data connection, datasets, visualization modeling, and other modules is shown in the following diagram:

In the top bar of the product interface, click "Data Preparation". The dropdown menu includes: Data Connection, Visualization Modeling, and Datasets. Click any entry point to access the corresponding module. As shown in the following diagram:

• Data Connections Overview: Creating a data connection can be understood as creating a credential for connecting to a database and accessing data tables. By creating a data connection, you can quickly obtain table information from the database, create datasets, and then process or cleanse input tables during the visualization modeling process.
• Dataset Overview : A dataset is the foundation for storing and managing data in DataWind, and also serves as the basis for visualization queries and analysis. In simple terms, it can be regarded as a type of data table. Users need to create datasets from the source tables that need to be analyzed in order to perform subsequent modeling, visualization chart creation, dashboard creation, and other tasks. In the dataset module, you can create, edit, and manage datasets.
• Overview of Visual Modeling: refers to the data processing and modeling capabilities provided by DataWind. This capability is commonly known as ETL (Extract-Transform-Load) and can also be referred to as data modeling. As an intermediate step between data sources and visual presentation, visual modeling enables users to efficiently process initial datasets through an easy-to-use, intelligent drag-and-drop interface before performing visual queries and creating dashboards. This allows data to be cleaned, transformed, and loaded, resulting in datasets that are more conducive to business analysis. At the same time, this module also supports model training, algorithmic prediction, and more.

The above provides a brief introduction to the three modules. For more information, please refer to the individual documentation for each module.