Pandas At Set Float

Sometimes you don't need a whole new table. You need to fix one number in the one you've got. Pandas At Set Float finds a single cell by row and column labels and overwrites it with a float - then hands the edited DataFrame back. It's the write-side counterpart to Pandas At Float, and it's the smallest possible "edit one value" tool in the pack.

How it works

The mechanism is dataframe.at[row_index, column_label] = data. You describe the target with the usual four inputs - row label, column label, and a string/int type switch for each - and supply the new value via the data FLOAT input. The node returns the frame with the cell replaced.

And the footnote that matters: this modifies the input DataFrame in place. pandas .at assignment edits the existing object rather than copying it, and the author says so explicitly in the docs. In a graph, any node reading from the same DataFrame output can see the change depending on execution order. For a simple workflow it's a non-issue; for a sprawling one, be deliberate about where the setter sits.

Inputs and outputs

• dataframe - the frame to edit (DATAFRAME)
• row_index - the row's label (STRING)
• row_index_type - string or int
• column_label - the column's label (STRING)
• column_label_type - string or int
• data - the float to write, default 0 (FLOAT)
• Output: DATAFRAME

When you'd reach for it

Correcting a bad value in a loaded table without rebuilding it; writing a computed result back into a results frame; or rounding out the read-modify-write loop - read with Pandas At Float, transform in a math node, write back here. It's niche, but when you need it, you need it, because the alternative is reconstructing the whole DataFrame.

Installing it

Part of ComfyUI-Data-Analysis (Hide Inada / HowToSD.com). Manager: search Data Analysis → install → restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/HowToSD/ComfyUI-Data-Analysis.git
mv ComfyUI-Data-Analysis data-analysis
pip install -r data-analysis/requirements.txt

No GPU; pandas comes via requirements.txt.

Gotchas

Same family theme: wrong label type enum → KeyError; label-based (not positional), so "third row, second column" wants Pandas Iat Set Float instead. And keep the in-place mutation in mind - if the same frame feeds other branches, place this node at the end of the line or accept that everything downstream sees the patched value.