The explorer lets you inspect a dataset and reshape it without writing code or touching the original file. Transforms are applied as a pipeline of steps, so you can always see and adjust what was done to the data.
Opening the explorer
Open a dataset in the data hub and choose Explore, or start from the data explore page and pick a dataset there. The explorer shows a preview of rows and columns that updates as you add transform steps.
The transforms available
- Rename or drop columns to tidy the table.
- Filter rows by text values, numeric ranges, or date ranges.
- Fill missing values, for example with zero, the column mean, or a placeholder label.
- Aggregate: group rows and summarise them, the classic group-and-count or group-and-average.
- Computed columns from simple arithmetic on existing columns, using plus, minus, multiply, and divide.
- Sort rows and remove duplicate rows.
Dataset transforms vs view transforms
Steps can be attached at two levels. Dataset-level transforms travel with the dataset, so every chart and exploration built on it starts from the cleaned shape. View-level transforms belong only to your current exploration, which is the right place for one-off filters. The original uploaded data is never modified either way.
Ready to visualise? The shape you build in the explorer feeds straight into charts; see Build charts.