Pandas Head

Pandas Head is the df.head() of the ComfyUI world: take a DataFrame, keep the first N rows, pass the rest along. In a notebook you'd type df.head() out of reflex to make sure your data actually loaded. In ComfyUI there's no console, so this node is the closest thing to that reflex - a way to trim a frame down to a manageable slice.

Two distinct jobs, and it's worth not confusing them. Job one: shrinking. You loaded a giant CSV and you only want the first handful of rows to, say, plot a sample. Set rows to 10 and downstream nodes see a 10-row frame. Job two: verifying. Feed the output into Pandas Show DataFrame and you get a readable table of what your data actually looks like - column names, types, whether that first row is garbage. The rows widget caps at 50, which is fine for eyeballing and not enough for real analysis, so treat this as a preview tool, not a data-splitting tool.

How it works

The node calls dataframe.head(rows) and returns the result. That's pandas' native head: the first rows rows, in order, with labels intact. It never mutates the input - you get a new, smaller frame, and the original is untouched, which is what you want when you're branching a workflow into "preview" and "full data" paths from one load node. Set rows to 0 and you get an empty frame, which is occasionally a useful way to stub something out.

Where people trip: head takes the first rows of whatever order the frame is in. If your CSV is unsorted, "first 10" tells you nothing about the distribution - it just shows you the top of the file. For a representative sample you'd want something like Pandas Value Counts or a shuffle first. Head is a glance at the beginning of the data, not a sample.

Inputs and outputs

• dataframe (required, DATAFRAME) - the frame to trim.
• rows (required, INT, default 10, max 50) - how many rows to keep.
• DATAFRAME output - the first rows rows.

Installing it

One pack, one install. ComfyUI Manager → search "Data analysis" → install ComfyUI-Data-Analysis → restart → reload the browser. Manual:

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

Requirements: pandas, matplotlib, seaborn, scipy, scikit-learn, openpyxl, lxml. No GPU, no model downloads. Head is one of the safest nodes in the pack - a one-line pandas wrapper - so if it's erroring, the frame feeding it is the problem, not the node. Check the column names and shape with the pack's introspection nodes before you trim.