- Print
- PDF
Creating and managing dataset
- Print
- PDF
Available in Classic and VPC
Creating and Managing Datasets guides you through precautions when constructing datasets and explains how to create, edit, upload, and delete datasets.
Precautions for constructing datasets
AiTEMS provides customized recommendation services based on dataset learning, so it is very important to configure the dataset.
The following are the precautions for constructing datasets.
Dataset types are user, item, and interaction, and learning is performed based on 3 types of datasets. Therefore, it is necessary to configure a dataset for each type.
The following describes each dataset type.- user: metadata containing user information (age, gender, etc.)
- item: metadata containing information related to the item (price, release date, category, etc.)
- interaction: metadata containing records obtained from interactions between users and items.
When constructing a dataset, you must configure the same fields as required schema fields.
Dataset type Required fields (Not NULL) user USER_ID (string) item ITEM_ID (string) interaction USER_ID (string)
ITEM_ID (string)
TIMESTAMP (long)Dataset fields must match schema fields not only in field names, but also in order of fields and case of field names.
You must not enter duplicate data values in the required fields of the user dataset and item dataset.
Create dataset
Create a dataset to use for learning. You can manage based on the dataset name, and a unique dataset ID is assigned when a dataset is created.
- To proceed with learning, you must create all datasets for each type.
- To create a dataset, you must have a dataset file ready for upload. For more information on configuring dataset files, see Precautions for constructing datasets.
- Only files in csv or csv.gz format can be used as dataset files.
The following describes how to create a dataset.
- Click the environment you are using in the Region menu and Platform menu of the NAVER Cloud Platform console.
- Click the Services > AI Services > AiTEMS menus in that order.
- Click the Dataset menu.
- Click the [Create dataset] button.
- When the screen to create dataset appears, enter the dataset name and set the dataset information.
- Dataset name: this is the criterion for managing the dataset. Enter between 5 and 20 characters.
- It must start with English and only English, numbers, '_', '-' can be entered.
- Description: enter the description of dataset
- Dataset type: select the type of dataset (user/item/interaction)
- For more details on the type of datasets, see Precautions for constructing datasets.
- After creating a dataset, you cannot edit the dataset type.
- Select schema: select a schema that is a required component of a dataset or create a new one.
- If no schema has been created, you can create a new schema by selecting Create New Schema.
- If you select Create New Schema, an area for configuring new schema fields appears.
- Select data: select a method for selecting a dataset file for learning
- Select from Object Storage: select if dataset files are uploaded to AiTEMS bucket of Object Storage
- Select from file: select if the dataset file is not uploaded to the AiTEMS bucket of Object Storage, and automatically save to the bucket if you select and upload the file.
- Dataset name: this is the criterion for managing the dataset. Enter between 5 and 20 characters.
- If you selected Create new schema in Select schema, click the [Add] button after setting the schema information.
- Schema name: this is the criterion for managing the schema. Enter between 3 and 20 characters.
- Schema description: enter the description of schema
- Field name: enter the same as the field name in the dataset file
- Field type: select the type of data to enter (string/float/long/double/int/boolean/null)
- Categorical field status: set to Y when entering data with a category
- You can change the order of the fields by dragging and dropping .
- If Select from Object Storage is selected in Select Data, click to select the dataset to upload.
- Only files uploaded to the AiTEMS bucket can be selected.
- When Select from file is selected in Select Data, either drag and drop the file to the Drag the file with your mouse or click here area, or click the applicable area to select the file.
- The selected files will be automatically uploaded to the path shown in Buckets/Paths.
- Click the [Create] button.
- From the notification pop-up window, click the [OK] button.
- A dataset is created and added to the list of datasets.
Edit dataset
You can edit a dataset's description or change its schema and dataset files.
The dataset type selected when creating the dataset cannot be modified.
The following describes how to edit a dataset.
- Click the environment you are using in the Region menu and Platform menu of the NAVER Cloud Platform console.
- Click the Services > AI Services > AiTEMS menus in that order.
- Click the Dataset menu.
- Click the desired dataset to edit in the list of datasets.
- When the pop-up window to edit the dataset appears, apply the items to be modified and then click the [Save and upload] button.
- To change the dataset file, click the [Change path and upload] button in the data edit pop-up window, click to select the file to change, and then click the [Change path and upload] button.
- You can also change the dataset file by clicking the [Upload dataset] button on the dataset screen. (See Upload dataset)
- From the notification pop-up window, click the [OK] button.
- If you change the dataset file, it changes to a pending state, and the state changes depending on whether the dataset file upload is successful or not.
Upload dataset
Enable to change dataset files.
- To change a dataset file, the desired file to change must be stored in an AiTEMS bucket in Object Storage.
- You can also change the dataset file through Edit dataset.
The following describes how to upload a dataset.
- Click the environment you are using in the Region menu and Platform menu of the NAVER Cloud Platform console.
- Click the Services > AI Services > AiTEMS menus in that order.
- Click the Dataset menu.
- Click the dataset you want to upload the dataset file to from the dataset list.
- Click the [Upload dataset] button.
- In the pop-up window to upload datasets, click to select the dataset file to be uploaded, and then click the [Request to upload] button.
- From the notification pop-up window, click the [OK] button.
- It changes to a pending state, and the state changes depending on whether the dataset file upload is successful or not.
Delete dataset
The following describes how to delete a dataset.
You cannot delete datasets connected to services. Delete the service first or change the dataset linked to the service before deleting the dataset.
- Click the environment you are using in the Region menu and Platform menu of the NAVER Cloud Platform console.
- Click the Services > AI Services > AiTEMS menus in that order.
- Click the Dataset menu.
- Click the desired dataset to delete in the list of datasets.
- Click the [Delete] button.
- When the pop-up window to delete dataset appears, enter the dataset name to delete and click the [Delete] button.
- The dataset is deleted from the list and is no longer visible.