Field survey · aggregated & modeled
The companion view to the case study catalog: every logged case study is tagged by discipline, tool, activity theme, year, and region — this page cross-tabulates those tags, tracks them over time, and fits a small logistic-regression model so you can ask "given X, what's likely?" and get a real answer back.
Loading catalog data…
How the mix of tags has shifted across 2023–2026. Bars are stacked counts; switch to share to normalize away the catalog's overall growth. Unspecified groups entries whose source page didn't state a year.
A few headline findings, pulled automatically from the data above.
Pick two tag dimensions to cross-tabulate. Each heatmap row is normalized to 100% — darker cells mean a larger share of that row's case studies carry that column's tag. The ranked list surfaces the strongest pairings across all dimensions, not just the two selected, weighted by how much they beat the catalog-wide baseline (lift), with a minimum of 5 case studies behind each claim.
A multinomial logistic regression, trained in your browser on the — catalogued entries. Choose what to predict and which tags to predict it from, then set a scenario below to see the model's probability distribution. With this little data the model is suggestive, not validated — read it as a second, model-based way of asking "what predicts what," alongside the lift ranking above, not as a fitted result you'd publish.