The features provided by the dataset are broadly divided into two categories: Data Preprocessing and Charts.
First, let's explain how to create a dataset:
Creating a Dataset
In data connections, select a data table and set appropriate conditions (such as filtering, joining, etc.) to create a core component of the SENSE system: a dataset. Subsequent data exploration and data management are all based on datasets.
The Sense system provides six ways to create a dataset:
- File Upload. The Sense system supports creating datasets by directly uploading files in CSV, XLSX, XLS, and BLF formats. All users with the
data analysis role can use this feature to create datasets directly. When creating a local file dataset, you can upload the file to the data source. Data sources that support file uploads include engine connections (built-in data connections), Greenplum, PostgreSQL, and Amazon Redshift. However, an existing dataset from the same source must be established in the application before local files can be uploaded to the corresponding non-built-in data sources.
- Data Connection. This feature can connect to various relational databases in the enterprise, such as Oracle, SQL Server, MySQL, etc.; NoSQL databases, such as Elastic Search, Solr, MongoDB, etc.; and big data platforms, such as Hive, Impala, etc., and then use the visual graphical operation interface to select the appropriate subset of data to create.
- SQL Query. On the basis of a data connection, datasets are created through custom SQL statements. This is suitable for users who have a deeper understanding of the SQL language.
- Multi-table Join. The SENSE system provides a powerful FUSION feature.
- Data Aggregation. On the basis of an existing dataset, datasets are created by aggregation, suitable for "data analysis" on datasets with many columns.
- Data Merge. When the data from multiple datasets need to be combined into one dataset, the data merge function can be used.