Add a dataset and inspect its data

A dataset is the tabular data you analyze. One collection can hold multiple datasets — this lesson registers your first one, by hand.

Upload a CSV to create a dataset

In the detail screen of the collection you made in Lesson 2, click Add item in the header and select Quick add….

• Drag the CSV file into the upload area, or click the area and select the file.
• A table preview appears as soon as you select it. Check that the values in the first row appear as the correct column names.
• If a column name is empty or invalid, fix it in the preview before uploading.

Click Upload 1 file(s) at the lower right. When the upload finishes, a new dataset named after the CSV file appears in the collection.

Tour the dataset — two tabs

Click the dataset you just made to open the detail screen. Two tabs to know.

Data tab

The uploaded content renders as a table. This is the first place to answer, “What shape did my data actually arrive in?” If you see unexpected empty cells or shifted columns, check the headers and values in the original CSV.

Schema tab

You see each column's name, alias, description, data type, and nullable setting. CSV uploads infer names and types automatically.

For existing columns that already contain uploaded data, Edit only unlocks the alias and description. The column name, data type, and nullable setting remain locked. A column added during editing allows all five fields to be configured. This is separate from correcting a CSV header—the column name—in the Quick add preview before upload.

Confirm three things.

1. Column names are English snake_case — not required for the system to work, but it'll be easier in the Analyst Path and Data Engineer Path code modes.
2. Type inference matches intent — especially price-style numeric columns mis-inferred as string will block aggregation and charting downstream.
3. Date-like columns have the intended type. A value such as 2026-07-01 can be inferred as Text.

If an existing column's name or type was inferred incorrectly, you cannot change it in place on this tab. Correct the header in the preview and upload again, define the schema before creating a new dataset, or produce a separate transformed output dataset with the required types.

Call global search once

From anywhere, press ⌘KCtrl+K⌘K / Ctrl+K to open global search. Type part of the dataset's name or alias. If the dataset came from a CSV, start with the filename without its extension.

Confirm that the result type is Dataset, then select it to open the detail screen. Global search finds resources without making you expand the collection tree manually. To find a column, open the dataset first and use its Schema tab; dataset column names are not part of global search.

What you should be able to do after this lesson

• The 3-click flow of CSV upload → dataset registration
• Inspect your data's shape via the Data and Schema tabs
• The global search shortcut (⌘KCtrl+K⌘K / Ctrl+K)

Next lesson

Build a first chart with one widget on top of the dataset, and pin it to a dashboard.