You can create a dataset on the Datasets tab of the administrator panel.
You'll need to give it a unique name, and set an access level (public, restricted via permission group, or shared directly with users) in order to save it.
The Overview tab of this page represents what the user will see on the published page. Initially the page is unpublished, so any changes you make and save will not be visible on your homepage.
Here you can add defining characteristics to help users locate your dataset and evaluate if it is something they'd like to work with. You can learn more about assigning general characteristics here.
You can also set a dataset as featured which means it will be displayed on your organization's homepage. If more than 3 datasets are set to featured, 3 will be randomly chosen when the page is loaded.
Populate your dataset with data so users can query it, use it in a project, and download it (if your permission system allows for it). Get started by creating a new table and uploading data to it.
When a dataset is created, it starts unpublished and with an in progress version. You can make as many updates and edits to this page as you'd like and it won't appear on your organization's homepage.
You can't publish this dataset without releasing the first version. When you're ready for your dataset to be discoverable, save any pending changes, and then click the Release and publish button.
You can unpublish this dataset at any time. If any users have it in their projects, its data and any derivative tables containing its data will be hidden from view and use.