SM
Dataset Resizer
Subset of Explore · ML Lab
Back to ML Lab
{% with msgs = get_flashed_messages(with_categories=true) %} {% if msgs %}
    {% for category, msg in msgs %}
  • {{ msg }}
    • {% endfor %}
{% endif %} {% endwith %}
Resize a dataset Explore subset

Upload any CSV that feels too large to work with. Choose the percentage you want to keep, and the resizer will return a smaller sample that preserves the overall feel of the data.

This is independent of the datasets listed in the ML Lab sidebar. You can bring in any CSV here.

We will sample roughly this share of rows. For very large files, even 5–10 % can be enough to explore patterns.

Show advanced options

If you supply a valid column name (for example a target label), the resizer will allocate rows per class so proportions remain close to the original.

Resize summary
{% if resize_info %}

Here is a quick view of what was produced from your last run. You can download the resized CSV and pass it straight into your modelling workflow.

  • Original rows: {{ resize_info.rows_in }}
  • Resized rows: {{ resize_info.rows_out }}
  • Target size: {{ "%.1f"|format(resize_info.pct) }} %
    • {% if resize_info.used_strat and resize_info.strat_col %}
  • Stratified by: {{ resize_info.strat_col }}
    • {% endif %}
Download resized CSV {% else %}

No resized dataset yet

Once you upload a file and choose a percentage, this panel will display the row counts and a download button for the resized sample.

{% endif %}