int_calculate_statistics(mki-整数计算统计)

ComfyUI's math nodes do arithmetic, but arithmetic isn't statistics. If you've got a pile of integers - a batch of seeds, strengths, step counts, whatever - and you want their mean, median, variance, or a z-score, int_calculate_statistics_makki (display name "int_calculate_statistics(mki-整数计算统计)") computes 21 different statistical measures over as many integer inputs as you care to wire in.

How it works

The node takes dynamic integer inputs - connect one int and another input slot appears, same pattern as the pack's dynamic-image nodes - and then, based on the stat_mode you pick, computes one measure over the whole set. The full menu:

• Basics: count, sum, mean, min, max, range
• Order stats: median, quartiles, iqr, mode, sorted_list, unique_values, cumulative_sum
• Spread: variance, std_dev, mad (mean absolute deviation)
• Shape: skewness, kurtosis (excess kurtosis, so a normal distribution reads 0)
• Relational: frequency, proportions, z_scores

The single output is typed * because what comes out depends on the mode: count returns one integer, variance one float, quartiles a list of three, z_scores a list of one per input, frequency a dict. Wire it into whatever consumer matches the shape.

Where it earns its keep

Honestly? Debugging and scheduling. People use it to sanity-check a batch of generated seeds (mean/range tells you if your "random" values are actually covering the space), to summarize a set of metrics, or to feed a single representative value downstream - e.g., compute the mean of several measured values and hand that to a strength control. The median is handy when one outlier would wreck a mean. It's a utility node, not a workhorse, but when you need a quick summary it's the difference between five math nodes and one.

Gotchas

• Empty input is handled, but not gracefully everywhere. With no integers connected, count returns 0; every other mode returns None. So if you run it unconnected you get None flowing into whatever you wired it to.
• proportions divides by the sum - if your integers sum to zero you'll get a division blow-up. Keep inputs nonzero for that mode.
• mode returns a list, not a single value - multiple values tie for most frequent and you get all of them. Don't wire it somewhere expecting one number.
• z_scores on a set with zero variance returns all zeros rather than crashing. That's a deliberate guard; it just means "everything is the mean," which is technically true.

Installing

It's part of ComfyUI-MakkiTools:

cd ComfyUI/custom_nodes
git clone https://github.com/MakkiShizu/ComfyUI-MakkiTools
cd ComfyUI-MakkiTools
pip install -r requirements.txt

Restart ComfyUI. Pure Python math + collections.Counter under the hood, no dependencies. It won't replace a spreadsheet, and the output typing ("what comes out depends on the mode") can trip up strict workflows - but for a quick summary of a batch of integers inside the graph, it's genuinely convenient.