Skip to content

Data Management

Load your own data, open a public dataset, or try the bundled demo.

Data Management: datasets and the tables of the selected dataset

Choose a dataset

Pick a dataset in Select dataset above the plot, or in the dataset list on the Data Management page:

  • Bundled demo: 24 trials from OneStop, with two participants and 12 texts.
  • Hand-drawn sample: a scanpath drawn by hand over a short English text, which you can edit.
  • Public datasets: OneStop and PoTeC. In the desktop app or a pip install, each downloads the first time you open it.
  • + → Import files: your own tables.
  • + → Draw a scanpath: write a text and place its fixations by hand.

OneStop

The bundled demo comes from OneStop Eye Movements (Berzak et al., 2025), an English corpus with 360 participants. The full corpus is available as four public datasets: OneStop · Ordinary reading, OneStop · Information seeking, OneStop · Repeated reading and OneStop · Information seeking (repeated). Each downloads once from OSF.

What your data needs

A dataset needs at least one of the first three tables below. Add the others as you need them.

Table Contents
Fixations One row per fixation: trial ID, duration, and x/y or a word ID
Words (interest areas) One row per word: trial ID, word ID and word box
Raw gaze One row per gaze sample: trial ID, x and y
Participants One row per participant, with metadata
Trials One row per trial, with metadata
Texts One row per text, with metadata

The app detects columns from common eye-tracker exports. Check and correct the mapping when you import. To keep condition labels and other columns, pick them under Extra fields to keep. You can filter, sort and group trials by any metadata field.

Files and units

  • CSV, TSV, TXT, Parquet, Feather, Excel, or a ZIP of these files
  • Positions in pixels
  • Times in milliseconds. A time column whose header names its unit, such as [s], [μs] or [ns], is converted.
  • Word boxes as x/y/width/height or EyeLink's IA_LEFT/RIGHT/TOP/BOTTOM
  • Word IDs as numbers, or as text such as w1. Spell text IDs the same way in the Fixations and Words tables. With text IDs, the order of the rows in the Words table sets the word order.

Trials over several screens

Map the Screen ID field in your tables. A column named screen_index sets the order of the screens. Without one, the screens are ordered by where they first appear. The app draws one screen at a time.

Reading measures

Corpus Analysis uses the per-word reading measures in your Words table, such as first fixation duration or total fixation duration. Map them when you import. EyeLink's IA_* columns are detected.

Add your own data

  1. Choose + → Import files and name the dataset.
  2. Upload your tables and check the column mapping.
  3. Under Recording setup, enter the Screen, Monitor width and Text size if you know them.
  4. Optionally, in the desktop app or a local install, attach the stimulus images from a folder.
  5. Click Add dataset, then Open in Scanpath to see its first trial. Later, Open in Scanpath beside Add dataset on the Data page does the same for the selected dataset.

Load and inspect in code

Load the same files in a script. The loaders detect column names the same way the app does.

Python and CLI 
import scanpath_studio as sps

words, fixations = sps.load_scanpath_data("ia.csv", "fixations.csv")
print(sps.list_trials(words, fixations))

scanpath-studio render --words ia.csv --fixations fixations.csv --list-trials
scanpath-studio check --words ia.csv --fixations fixations.csv

For raw gaze, metadata and explicit column mappings, see the loaders and CLI input options.

Manage datasets

Click a dataset in the dataset list to open it. Below the list, What's in ‹dataset› shows its tables, annotations and Data checks. When a check flags more than half of a table, such as every duration read as 0 ms, the Scanpath view says so above the plot.

Use Edit dataset to change its name, column mapping, recording setup or metadata, then click Save changes. Download setup file saves the mapping and recording setup, so you can apply them to similar data.

The editor offers only the columns kept when the dataset was added. To use a column that isn't listed, remove the dataset and add it again from its files, mapping the column or keeping it under Extra fields to keep.