type audit script, fix misclassified activities, alpineski excluded from ascent

scripts/audit-types.js

@@ -0,0 +1,138 @@
+#!/usr/bin/env node
+// Audit activity types against their measured stats and fix contradictions.
+//
+//   node scripts/audit-types.js          # dry run: show proposed changes
+//   node scripts/audit-types.js --apply  # apply + log to data/type-fixes.json
+//
+// Custom/curated types (skitour, vita, iceskate, …) are left alone; only
+// clear physics contradictions in the generic foot/bike types plus the
+// "outdoor" fallback are reclassified, using centroids computed from the
+// database's own labelled activities.
+
+import Database from 'better-sqlite3';
+import fs from 'node:fs';
+
+const APPLY = process.argv.includes('--apply');
+const dbPath = (process.env.STREBA_DATA_DIR ?? 'data') + '/streba.db';
+const db = new Database(dbPath);
+
+const rows = db
+	.prepare(
+		`SELECT id, name, type, distance_m, elev_gain_m,
+		        coalesce(moving_s, duration_s) time_s
+		 FROM activities`
+	)
+	.all()
+	.map((r) => ({
+		...r,
+		kmh: r.time_s > 0 ? (r.distance_m / r.time_s) * 3.6 : null,
+		gainPerKm: r.distance_m > 0 ? r.elev_gain_m / (r.distance_m / 1000) : 0
+	}));
+
+// generic classes we are willing to assign
+const FOOT_SLOW = new Set(['hike', 'walk', 'walking']);
+const FOOT_FAST = new Set(['run', 'running', 'trail run']);
+const BIKE = new Set(['ride', 'cycling', 'bike', 'mtb', 'gravel']);
+const GENERIC = new Set([...FOOT_SLOW, ...FOOT_FAST, ...BIKE, 'outdoor']);
+
+// centroids from the user's own labelled data (fall back to sensible defaults)
+function centroid(types, defKmh, defGain, defKm) {
+	const sample = rows.filter((r) => types.has(r.type) && r.kmh !== null);
+	if (sample.length < 3) return { kmh: defKmh, gainPerKm: defGain, km: defKm };
+	return {
+		kmh: sample.reduce((s, r) => s + r.kmh, 0) / sample.length,
+		gainPerKm: sample.reduce((s, r) => s + r.gainPerKm, 0) / sample.length,
+		km: sample.reduce((s, r) => s + r.distance_m / 1000, 0) / sample.length
+	};
+}
+const CLASSES = [
+	{ name: 'hike', c: centroid(FOOT_SLOW, 4, 70, 10) },
+	{ name: 'run', c: centroid(FOOT_FAST, 11, 25, 8) },
+	{ name: 'ride', c: centroid(new Set(['ride', 'cycling', 'bike']), 21, 15, 40) },
+	{ name: 'mtb', c: centroid(new Set(['mtb']), 13.5, 45, 27) }
+];
+function classify(r) {
+	if (r.kmh === null) return null;
+	let best = null;
+	let bestScore = Infinity;
+	for (const cls of CLASSES) {
+		// normalised distance in (speed, climb-rate, length) space
+		const ds = (r.kmh - cls.c.kmh) / 6;
+		const dg = (r.gainPerKm - cls.c.gainPerKm) / 40;
+		const dd = (r.distance_m / 1000 - cls.c.km) / 20;
+		const score = ds * ds + dg * dg + dd * dd;
+		if (score < bestScore) {
+			bestScore = score;
+			best = cls.name;
+		}
+	}
+	return best;
+}
+
+function contradiction(r) {
+	if (r.kmh === null) return false;
+	if (r.type === 'outdoor') return true;
+	if (/^\d+$/.test(r.type)) return true;
+	if (FOOT_SLOW.has(r.type) && r.kmh > 9) return true;
+	if (FOOT_FAST.has(r.type) && (r.kmh < 4.5 || r.kmh > 16)) return true;
+	if (BIKE.has(r.type) && r.kmh < 5.5) return true;
+	return false;
+}
+
+const changes = [];
+for (const r of rows) {
+	if (!GENERIC.has(r.type) && !/^\d+$/.test(r.type)) continue; // curated type: hands off
+	if (!contradiction(r)) continue;
+	const suggestion = classify(r);
+	if (!suggestion || suggestion === r.type) continue;
+	changes.push({
+		id: r.id,
+		name: r.name,
+		from: r.type,
+		to: suggestion,
+		kmh: r.kmh === null ? null : Math.round(r.kmh * 10) / 10,
+		gainPerKm: Math.round(r.gainPerKm)
+	});
+}
+
+// borderline cases worth a human look (reported, never auto-changed)
+const review = rows.filter(
+	(r) =>
+		(r.type === 'ride' && r.gainPerKm > 35 && r.kmh < 16) ||
+		(r.type === 'mtb' && r.gainPerKm < 12 && r.kmh > 17) ||
+		(FOOT_FAST.has(r.type) && r.kmh !== null && r.kmh < 6) ||
+		(FOOT_SLOW.has(r.type) && r.kmh !== null && r.kmh > 7 && r.kmh <= 9)
+);
+if (review.length > 0) {
+	console.log('\nBorderline (left unchanged - review by hand if they look off):');
+	console.table(
+		review.map((r) => ({
+			id: r.id,
+			name: r.name,
+			type: r.type,
+			kmh: Math.round((r.kmh ?? 0) * 10) / 10,
+			gainPerKm: Math.round(r.gainPerKm),
+			km: Math.round(r.distance_m / 100) / 10
+		}))
+	);
+}
+
+if (changes.length === 0) {
+	console.log('No contradictions found.');
+	process.exit(0);
+}
+console.table(changes);
+if (!APPLY) {
+	console.log(`Dry run - ${changes.length} change(s) proposed. Re-run with --apply.`);
+	process.exit(0);
+}
+
+const update = db.prepare('UPDATE activities SET type = ? WHERE id = ?');
+db.transaction(() => {
+	for (const change of changes) update.run(change.to, change.id);
+})();
+fs.writeFileSync(
+	(process.env.STREBA_DATA_DIR ?? 'data') + '/type-fixes.json',
+	JSON.stringify({ applied: new Date().toISOString(), changes }, null, 2)
+);
+console.log(`Applied ${changes.length} change(s); log written to data/type-fixes.json.`);

src/lib/activity-colors.ts

@@ -31,10 +31,21 @@ const TYPE_SLOT: Record<string, number> = {
 	'via ferrata': 3,
 	alpinism: 3,
 	ski: 6,
+	alpineski: 6,
+	skitour: 6,
+	langlauf: 6,
 	'backcountry ski': 6,
 	'nordic ski': 6,
 	snowboard: 6,
 	snowshoe: 6,
+	mtb: 2,
+	gravel: 2,
+	vita: 1,
+	iceskate: 4,
+	inlineskate: 4,
+	inline: 4,
+	'hike & fly': 3,
+	watersport: 5,
 	segment: 7
 };
 

src/lib/activity-rules.ts

@@ -1,7 +1,13 @@
 // Lift-assisted downhill types: their recorded elevation gain comes from
 // lifts, so it doesn't count toward ascent totals. Touring variants
-// ("ski touring", "backcountry ski") earn their metres and do count.
-export const LIFT_ASSISTED_TYPES = ['ski', 'alpine ski', 'downhill ski', 'snowboard'];
+// ("skitour", "backcountry ski", "langlauf") earn their metres and do count.
+export const LIFT_ASSISTED_TYPES = [
+	'ski',
+	'alpine ski',
+	'alpineski',
+	'downhill ski',
+	'snowboard'
+];
 
 export function countsTowardAscent(type: string): boolean {
 	return !LIFT_ASSISTED_TYPES.includes(type.toLowerCase().trim());