Most analysis errors don't happen in the model. They happen earlier — in a merge that silently dropped rows, an exclusion rule applied inconsistently, a sensor artifact that became a "finding." Neutropic treats data preparation as the part of analysis that most needs daylight.
Load once, reference forever
Large files — survey exports, interaction logs, video, physiological recordings — are loaded once per session and referenced by name afterwards. No re-uploading, no guessing which of three near-identical CSVs is current: the session knows its data, and every artifact records which snapshot it came from.
Cleaning that keeps a log
Every preparation step is explicit and recorded:
- Missing data is reported per variable and per participant — with patterns flagged — before any imputation or exclusion is applied.
- Exclusion rules are stated as rules, applied uniformly, and counted, so the flow from raw N to analyzed N is a table, not a mystery.
- Merges and reshapes report what matched and what didn't, instead of silently intersecting.
The result is a cleaning log you can paste into a methods section — because it is one.
Multimodal, on one timeline
Facial action units, gaze, EDA, and HRV load as first-class data alongside the survey and behavioral files, aligned to a shared timeline using your event markers. The alignment is shown before analysis, so an offset never becomes an effect.
Exploration before confirmation
Distributions, outliers, and anomalies are surfaced visually per condition before any hypothesis test runs. When something looks wrong — a ceiling effect, a bimodal response pattern, a flat sensor channel — you find out now, not from a reviewer.
