Data & Research
Does Editing Pace Change How Grateful Viewers Feel? (Ongoing)
An ongoing study testing whether a DIY video's editing pace changes how viewers respond in the comments — 11K videos, 3M+ comments, still in analysis.
Problem
Instructional/DIY content lives or dies on whether viewers feel the creator respected their time and effort. Does the pacing of a video — how much motion and scene-change happens per second — actually shift how grateful people are in the comments, or is that assumption untested folklore?
Approach
Built a pipeline across 708 DIY/instructional search queries → videos → frame- and scene-level optical-flow motion extraction → GPT-based comment classification (Gratitude / Not Gratitude) across 3M+ comments, then built an interactive dashboard to explore the full dataset (searchable, sortable, exportable across linked video, comment, and motion tables).
Results
Still in analysis — the pacing-vs-gratitude relationship hasn't been conclusively established yet. This entry will update once the analysis is complete.
Stack: Python · NLP · Computer Vision · GPT Classification