make-mp-data 3.0.1 → 3.0.2

package/dungeons/adspend.js

@@ -18,7 +18,7 @@ import { pickAWinner, weighNumRange, date, integer } from "../lib/utils/utils.js
 
 /** @type {import('../types').Dungeon} */
 const config = {
-	token: process.env.MASTER_PROJECT_TOKEN || "",
+	token: "",
 	seed: "foo bar",
 	numDays: 365, //how many days worth of data
 	numEvents: 10000000, //how many events
@@ -87,6 +87,40 @@ const config = {
 	groupProps: {},
 	lookupTables: [],
 	hook: function (record, type, meta) {
+		// user hook: segment users into tiers based on their lucky number
+		if (type === "user") {
+			if (record.luckyNumber > 300) {
+				record.tier = "gold";
+			} else if (record.luckyNumber > 150) {
+				record.tier = "silver";
+			} else {
+				record.tier = "bronze";
+			}
+		}
+
+		// event hook: gold tier users get a "premium" color override 20% of the time
+		if (type === "event") {
+			if (record.color === "violet" || record.color === "indigo") {
+				record.is_rare_color = true;
+				record.number = (record.number || 0) * 2;
+			}
+		}
+
+		// everything hook: users with many events get a bonus "milestone" event
+		if (type === "everything") {
+			if (record.length > 50) {
+				const lastEvent = record[record.length - 1];
+				record.push({
+					event: "milestone reached",
+					time: lastEvent.time,
+					user_id: lastEvent.user_id,
+					milestone: record.length,
+					color: "gold",
+				});
+			}
+			return record;
+		}
+
 		return record;
 	}
 };

package/dungeons/anon.js

@@ -20,7 +20,7 @@ import { pickAWinner, weighNumRange, date, integer } from "../lib/utils/utils.js
 
 /** @type {import('../types').Dungeon} */
 const config = {
-	token: process.env.MASTER_PROJECT_TOKEN || "",
+	token: "",
 	seed: "foo bar",
 	numDays: 365, //how many days worth of data
 	numEvents: 100000, //how many events
@@ -95,6 +95,30 @@ const config = {
 	groupProps: {},
 	lookupTables: [],
 	hook: function (record, type, meta) {
+		// user hook: assign engagement tier based on lucky number
+		if (type === "user") {
+			record.engagement_tier = record.luckyNumber > 250 ? "high" : "low";
+		}
+
+		// event hook: weight-1 events ("yak") are treated as error/rare signals
+		if (type === "event") {
+			if (record.event === "yak") {
+				record.is_error = true;
+			}
+			if (record.event === "foo") {
+				record.priority = "high";
+			}
+		}
+
+		// everything hook: low-activity anonymous users churn — keep only first 60% of events
+		if (type === "everything") {
+			if (record.length < 8) {
+				const cutoff = Math.ceil(record.length * 0.6);
+				return record.slice(0, cutoff);
+			}
+			return record;
+		}
+
 		return record;
 	}
 };

package/dungeons/{array-of-object-loopup.js → array-of-object-lookup.js}

@@ -19,7 +19,7 @@ const spiritAnimals = ["duck", "dog", "otter", "penguin", "cat", "elephant", "li
 
 /** @type {import('../types.js').Dungeon} */
 const config = {
-	// token: "0a3e6aa01225ed03856dca545c2b5b3d",
+	// token: "",
 	seed: "test array of objects lookup",
 	name: "array-of-object-lookup",
 	numDays: 60, //how many days worth1 of data
@@ -41,7 +41,7 @@ const config = {
 	alsoInferFunnels: true,
 	concurrency: 1,
 	batchSize: 250_000,
-	writeToDisk: true,
+	writeToDisk: false,
 	events: [
 		{
 			event: "checkout",
@@ -112,18 +112,38 @@ const config = {
 
 		const NOW = dayjs();
 
-
 		if (type === "event") {
-			
+			// Pattern 1: Checkouts with coupons get a discount_applied flag and adjusted total
+			if (record.event === "checkout" && record.coupon && record.coupon !== "none") {
+				record.discount_applied = true;
+				const pctMatch = record.coupon.match(/(\d+)%/);
+				if (pctMatch) {
+					record.discount_percent = parseInt(pctMatch[1]);
+				}
+			}
+
+			// Pattern 2: "save item" events on weekends are tagged as wishlist behavior
+			if (record.event === "save item") {
+				const dow = dayjs(record.time).day();
+				if (dow === 0 || dow === 6) {
+					record.save_context = "weekend_browse";
+				} else {
+					record.save_context = "weekday_intent";
+				}
+			}
 		}
 
 		if (type === "everything") {
-			
-
+			// Pattern 3: Users who view 5+ items but never checkout are tagged as window shoppers
+			const views = record.filter(e => e.event === "view item").length;
+			const checkouts = record.filter(e => e.event === "checkout").length;
+			if (views >= 5 && checkouts === 0) {
+				for (const e of record) {
+					e.user_segment = "window_shopper";
+				}
+			}
 		}
 
-
-
 		return record;
 	}
 };

package/dungeons/benchmark-heavy.js

@@ -107,7 +107,7 @@ const commentGenerator = createTextGenerator({
 
 /** @type {import('../types.js').Dungeon} */
 const config = {
-	// token: process.env.MASTER_PROJECT_TOKEN || "",
+	// token: "",
 	name: "300k-Events-Heavy",
 	format: 'json', //csv or json
 	seed: "one million events",
@@ -130,7 +130,7 @@ const config = {
 	isAnonymous: false,
 	alsoInferFunnels: true,
 	// concurrency automatically set to 1 when strictEventCount is enabled
-	writeToDisk: true,
+	writeToDisk: false,
 	batchSize: 2_500_000,
 
 	events: [

package/dungeons/benchmark-light.js

@@ -18,7 +18,7 @@ import { createTextGenerator } from '../lib/generators/text.js';
 
 /** @type {import('../types.js').Dungeon} */
 const config = {
-	// token: process.env.MASTER_PROJECT_TOKEN || "",
+	// token: "",
 	name: "5M-Events-Light",
 	format: 'json', //csv or json
 	seed: "one million events",
@@ -41,7 +41,7 @@ const config = {
 	isAnonymous: false,
 	alsoInferFunnels: true,
 	// concurrency automatically set to 1 when strictEventCount is enabled
-	writeToDisk: true,
+	writeToDisk: false,
 	batchSize: 12_500_000,
 
 	events: [

package/dungeons/big.js

@@ -19,7 +19,7 @@ const totalDays = (numQuarters * 90) + 10;
 
 /** @type {import('../types').Dungeon} */
 const config = {
-	// token: process.env.MASTER_PROJECT_TOKEN || "",
+	// token: "",
 	seed: seed,
 	numDays: totalDays,
 	numEvents: totalEvents,
@@ -39,7 +39,7 @@ const config = {
 	hasCampaigns: false,
 	hasDesktopDevices: false,
 	hasIOSDevices: false,
-	writeToDisk: "gs://dungeon_master_4/big_data",
+	writeToDisk: false,
 	funnels: [
 		{
 			"sequence": ["foo", "bar", "baz", "qux", "garply", "durtle", "linny", "fonk", "crumn", "yak"],

package/dungeons/business.js

@@ -1,20 +1,21 @@
 /**
- * This is the default configuration file for the data generator in COMPLEX mode
- * notice how the config object is structured, and see it's type definition in ./types.d.ts
- * feel free to modify this file to customize the data you generate
- * see helper functions in utils.js for more ways to generate data
+ * Video platform dungeon (business/complex mode)
  */
 
+import Chance from 'chance';
+import dayjs from 'dayjs';
+import utc from 'dayjs/plugin/utc.js';
+dayjs.extend(utc);
+import { weighNumRange, pickAWinner, exhaust } from '../lib/utils/utils.js';
+import * as v from 'ak-tools';
 
-const Chance = require('chance');
 const chance = new Chance();
-const { weighNumRange, date, integer, pickAWinner, exhaust } = require('../lib/utils/utils.js');
-const u = require('ak-tools');
+const integer = (min, max) => chance.integer({ min, max });
 
 const channel_ids = [...Array(1234).keys()].map(i => i + 1).map(n => `channel_id_${n}`);
-const channel_names = chance.n(u.makeName, 1234);
+const channel_names = chance.n(v.makeName, 1234);
 const video_ids = [...Array(50000).keys()].map(i => i + 1).map(n => n.toString());
-const video_names = chance.n(u.makeName, 50000);
+const video_names = chance.n(v.makeName, 50000);
 
 const EVENTS = 50_000
 const USERS = EVENTS / 100
@@ -22,7 +23,7 @@ const USERS = EVENTS / 100
 
 /** @type {import('../types.js').Dungeon} */
 const config = {
-	token: process.env.MASTER_PROJECT_TOKEN || "",
+	token: "",
 	seed: "it's business time...",
 	numDays: 90, //how many days worth of data
 	numEvents: EVENTS, //how many events
@@ -258,6 +259,52 @@ const config = {
 	],
 
 	hook: function (record, type, meta) {
+		// --- user hook: tag users by their experiment variant ---
+		if (type === "user") {
+			const exp = record.experiment;
+			if (exp && Array.isArray(exp) && exp[0]) {
+				record.experimentGroup = exp[0].variant || "unknown";
+			}
+			return record;
+		}
+
+		// --- event hook: weekend watch time boost + premium quality bias ---
+		if (type === "event") {
+			if (record.event === "watch video" && record.time) {
+				const day = dayjs(record.time).day();
+				if (day === 0 || day === 6) {
+					record["watch time (sec)"] = Math.round((record["watch time (sec)"] || 60) * 1.8);
+					record.is_weekend_session = true;
+				}
+			}
+			// app errors on weekends are more severe (skeleton crew)
+			if (record.event === "app error" && record.time) {
+				const day = dayjs(record.time).day();
+				if (day === 0 || day === 6) {
+					record.weekend_incident = true;
+				}
+			}
+			return record;
+		}
+
+		// --- everything hook: binge watchers (5+ watch events) get a "binge_session" event ---
+		if (type === "everything") {
+			const watches = record.filter(e => e.event === "watch video");
+			if (watches.length >= 5) {
+				const totalWatchTime = watches.reduce((sum, e) => sum + (e["watch time (sec)"] || 0), 0);
+				const lastWatch = watches[watches.length - 1];
+				record.push({
+					event: "binge_session",
+					time: dayjs(lastWatch.time).add(1, "minute").toISOString(),
+					user_id: lastWatch.user_id,
+					videos_watched: watches.length,
+					total_watch_time_sec: totalWatchTime,
+					avg_watch_time_sec: Math.round(totalWatchTime / watches.length)
+				});
+			}
+			return record;
+		}
+
 		return record;
 	}
 };
@@ -267,7 +314,7 @@ const config = {
 function makeHashTags() {
 	const possibleHashtags = [];
 	for (let i = 0; i < 20; i++) {
-		possibleHashtags.push('#' + u.makeName(2, ''));
+		possibleHashtags.push('#' + v.makeName(2, ''));
 	}
 
 	const numHashtags = integer(integer(1, 5), integer(5, 10));
@@ -342,4 +389,4 @@ function designExperiment() {
 
 
 
-module.exports = config;
+export default config;

package/dungeons/complex.js

@@ -8,12 +8,15 @@
 
 import Chance from 'chance';
 const chance = new Chance();
+import dayjs from "dayjs";
+import utc from "dayjs/plugin/utc.js";
+dayjs.extend(utc);
 import { weighNumRange, date, integer } from "../lib/utils/utils.js";
 import * as u from 'ak-tools';
 
 /** @type {import('../types.js').Dungeon} */
 const config = {
-	token: process.env.MASTER_PROJECT_TOKEN || "",
+	token: "",
 	seed: "quite complexus",
 	numDays: 30, //how many days worth of data
 	numEvents: 100_000, //how many events
@@ -281,6 +284,36 @@ const config = {
 	],
 
 	hook: function (record, type, meta) {
+		// event hook: weekend watch time boost — videos watched on weekends get 1.5x duration
+		if (type === "event") {
+			if (record.event === "watch video" && record.time) {
+				const day = dayjs(record.time).day();
+				if (day === 0 || day === 6) {
+					record.watchTimeSec = Math.round((record.watchTimeSec || 60) * 1.5);
+					record.is_weekend = true;
+				}
+			}
+			// support tickets on high-severity get escalated flag
+			if (record.event === "support ticket" && record.severity === "high") {
+				record.escalated = true;
+			}
+		}
+
+		// everything hook: simulate cart abandonment — users who "add to cart" but never "checkout" lose their last add-to-cart
+		if (type === "everything") {
+			const hasCheckout = record.some(e => e.event === "checkout");
+			const hasAddToCart = record.some(e => e.event === "add to cart");
+			if (hasAddToCart && !hasCheckout) {
+				// mark all their add-to-cart events as abandoned
+				for (const e of record) {
+					if (e.event === "add to cart") {
+						e.abandoned = true;
+					}
+				}
+			}
+			return record;
+		}
+
 		return record;
 	}
 };

package/dungeons/copilot.js

@@ -66,7 +66,7 @@ function makeProducts(maxItems = 5) {
 
 /** @type {import('../types').Dungeon} */
 const config = {
-	token: "a4152b30c5276306151de1146321997e",
+	token: "",
 	seed: "simple is best",
 	numDays: 108, //how many days worth1 of data
 	numEvents: 2_000_000, //how many events

package/dungeons/{harness/harness-education.js → education.js}

@@ -1,7 +1,7 @@
 import dayjs from "dayjs";
 import utc from "dayjs/plugin/utc.js";
 import "dotenv/config";
-import * as u from "../../lib/utils/utils.js";
+import * as u from "../lib/utils/utils.js";
 import * as v from "ak-tools";
 
 const SEED = "harness-education";
@@ -95,7 +95,7 @@ const problemIds = v.range(1, 601).map(n => `problem_${v.uid(6)}`);
 
 /** @type {Config} */
 const config = {
-	token: "10a2fd5d566edd19e803036b276fe91b",
+	token: "",
 	seed: SEED,
 	numDays: days,
 	numEvents: num_users * 120,
@@ -457,7 +457,7 @@ const config = {
 		}
 
 		// ═══════════════════════════════════════════════════════════════════
-		// Hook #6: SEMESTER-END SPIKE
+		// Hook #6: SEMESTER-END SPIKE (tag in event hook, duplicate in everything hook)
 		// ═══════════════════════════════════════════════════════════════════
 		if (type === "event") {
 			if (record.time) {
@@ -468,14 +468,6 @@ const config = {
 				if (spikableEvents.includes(record.event)) {
 					if (dayInDataset >= 75 && dayInDataset <= 85) {
 						record.semester_end_rush = true;
-
-						// 50% chance to duplicate
-						if (chance.bool({ likelihood: 50 })) {
-							const duplicate = JSON.parse(JSON.stringify(record));
-							duplicate.time = eventTime.add(chance.integer({ min: 5, max: 120 }), 'minutes').toISOString();
-							duplicate.semester_end_rush = true;
-							return [record, duplicate];
-						}
 					} else {
 						record.semester_end_rush = false;
 					}
@@ -628,6 +620,31 @@ const config = {
 					}
 				});
 			}
+
+			// Hook #6: SEMESTER-END SPIKE - duplicate assessment events in the spike window
+			const duplicates = [];
+			userEvents.forEach((event) => {
+				if (event.semester_end_rush === true && chance.bool({ likelihood: 50 })) {
+					const dup = JSON.parse(JSON.stringify(event));
+					dup.time = dayjs(event.time).add(chance.integer({ min: 5, max: 120 }), 'minutes').toISOString();
+					dup.semester_end_rush = true;
+					duplicates.push(dup);
+				}
+			});
+			if (duplicates.length > 0) {
+				userEvents.push(...duplicates);
+			}
+
+			// Hook #7: FREE VS PAID - reinforce the subscription effect on certificates
+			const subStatus = userEvents.length > 0 ? userEvents[0].subscription_status : "free";
+			if (subStatus === "free") {
+				// Free users lose 40% of their certificates (simulating lower completion)
+				for (let i = userEvents.length - 1; i >= 0; i--) {
+					if (userEvents[i].event === "certificate earned" && chance.bool({ likelihood: 40 })) {
+						userEvents.splice(i, 1);
+					}
+				}
+			}
 		}
 
 		// ═══════════════════════════════════════════════════════════════════
@@ -944,7 +961,7 @@ export default config;
  *   import config from './dungeons/harness-education.js';
  *   const results = await generate(config);
  *
- * OUTPUT FILES (with writeToDisk: true, format: "json", gzip: true):
+ * OUTPUT FILES (with writeToDisk: false, format: "json", gzip: true):
  *
  *   - needle-haystack-education__events.json.gz - All event data
  *   - needle-haystack-education__user_profiles.json.gz - User profiles

package/dungeons/experiments.js

@@ -15,7 +15,7 @@ const days = 100;
 
 /** @type {Config} */
 const config = {
-	token: process.env.MASTER_PROJECT_TOKEN || "",
+	token: "",
 	seed: SEED,
 	numDays: days,
 	numEvents: num_users * 100,
@@ -96,6 +96,12 @@ const config = {
 
 		if (type === "event") {
 			const EVENT_TIME = dayjs(record.time);
+			// Pattern 1: "variant 1" users spend more money (winning variant)
+			if (record.event === "money event" && record["Variant name"] === "variant 1") {
+				record.amount = Math.round((record.amount || 50) * 1.5);
+			}
+			// Pattern 2: "control" users see fewer rare events (simulating lower engagement)
+			// handled in "everything" below
 		}
 
 		if (type === "user") {
@@ -115,7 +121,14 @@ const config = {
 		}
 
 		if (type === "everything") {
-
+			// Pattern 3: Control variant users have 30% of their "rare event" events removed
+			// to simulate lower engagement in control group
+			return record.filter(e => {
+				if (e.event === "rare event" && e["Variant name"] === "control" && Math.random() < 0.3) {
+					return false;
+				}
+				return true;
+			});
 		}
 
 		return record;

package/dungeons/{harness/harness-fintech.js → fintech.js}

@@ -1,7 +1,7 @@
 import dayjs from "dayjs";
 import utc from "dayjs/plugin/utc.js";
 import "dotenv/config";
-import * as u from "../../lib/utils/utils.js";
+import * as u from "../lib/utils/utils.js";
 
 const SEED = "harness-fintech";
 dayjs.extend(utc);
@@ -83,7 +83,7 @@ const days = 100;
 
 /** @type {Config} */
 const config = {
-	token: "048f954dddbfbd4bff34edd8b3d95c06",
+	token: "",
 	seed: SEED,
 	numDays: days,
 	numEvents: num_users * 120,
@@ -192,7 +192,7 @@ const config = {
 			event: "balance checked",
 			weight: 15,
 			properties: {
-				"account_balance": u.weighNumRange(0, 50000, 0.3, 2500),
+				"account_balance": u.weighNumRange(0, 50000, 0.8, 2500),
 				"account_type": u.pickAWinner(["checking", "savings", "investment"]),
 			}
 		},
@@ -405,7 +405,7 @@ const config = {
 			// Payday: 1st and 15th
 			if (record.event === "transaction completed" && record.transaction_type === "direct_deposit") {
 				if (dayOfMonth === 1 || dayOfMonth === 15) {
-					record.amount = Math.floor((record.amount || 50) * 2);
+					record.amount = Math.floor((record.amount || 50) * 3);
 					record.payday = true;
 				} else {
 					record.payday = false;
@@ -415,8 +415,8 @@ const config = {
 			// Post-payday spending: days 1-3 and 15-17
 			if (record.event === "transfer sent") {
 				const isPaydayWindow = (dayOfMonth >= 1 && dayOfMonth <= 3) || (dayOfMonth >= 15 && dayOfMonth <= 17);
-				if (isPaydayWindow && chance.bool({ likelihood: 40 })) {
-					record.amount = Math.floor((record.amount || 200) * 1.5);
+				if (isPaydayWindow && chance.bool({ likelihood: 60 })) {
+					record.amount = Math.floor((record.amount || 200) * 2.0);
 					record.post_payday_spending = true;
 				} else {
 					record.post_payday_spending = false;
@@ -579,7 +579,7 @@ const config = {
 			// -----------------------------------------------------------
 			let lowBalanceChecks = 0;
 			userEvents.forEach((event) => {
-				if (event.event === "balance checked" && (event.account_balance || 0) < 500) {
+				if (event.event === "balance checked" && (event.account_balance || 0) < 3000) {
 					lowBalanceChecks++;
 				}
 			});
@@ -934,7 +934,7 @@ export default config;
  *   import config from './dungeons/harness-fintech.js';
  *   const results = await generate(config);
  *
- * OUTPUT FILES (with writeToDisk: true):
+ * OUTPUT FILES (with writeToDisk: false):
  *
  *   - needle-haystack-fintech__events.json.gz       - All event data
  *   - needle-haystack-fintech__user_profiles.json.gz - User profiles

package/dungeons/foobar.js

@@ -136,7 +136,7 @@ const seed = Math.random().toString()
 
 /** @type {import('../types').Dungeon} */
 const config = {
-	token: process.env.MASTER_PROJECT_TOKEN || "",
+	token: "",
 	seed: seed,
 	numDays: 30, //how many days worth of data
 	numEvents: numEvents, //how many events
@@ -232,6 +232,38 @@ const config = {
 	},
 
 	hook: function (record, type, meta) {
+		// event hook: high-weight events get a "hot" flag, low-weight get "cold"
+		if (type === "event") {
+			const hotEvents = ["foo", "bar", "baz"];
+			const coldEvents = ["crumn", "yak"];
+			if (hotEvents.includes(record.event)) {
+				record.temperature = "hot";
+			} else if (coldEvents.includes(record.event)) {
+				record.temperature = "cold";
+			} else {
+				record.temperature = "warm";
+			}
+		}
+
+		// everything hook: hash-based cohort — 10% of users (by distinct_id) get doubled events
+		if (type === "everything") {
+			if (record.length > 0) {
+				const userId = record[0].user_id || record[0].distinct_id || "";
+				if (userId && userId.charCodeAt(0) % 10 === 0) {
+					// power user: duplicate each event with a slight time offset
+					const extras = record.slice(0, 3).map(e => ({
+						...e,
+						event: e.event,
+						user_id: e.user_id,
+						time: e.time,
+						is_duplicate: true,
+					}));
+					return record.concat(extras);
+				}
+			}
+			return record;
+		}
+
 		return record;
 	}
 };

package/dungeons/{harness/harness-food.js → food.js}

@@ -1,7 +1,7 @@
 import dayjs from "dayjs";
 import utc from "dayjs/plugin/utc.js";
 import "dotenv/config";
-import * as u from "../../lib/utils/utils.js";
+import * as u from "../lib/utils/utils.js";
 import * as v from "ak-tools";
 
 const SEED = "harness-food";
@@ -77,7 +77,7 @@ const couponCodes = v.range(1, 51).map(n => `QUICK${v.uid(5).toUpperCase()}`);
 
 /** @type {Config} */
 const config = {
-	token: "9ffdb0649ebbea4ff7a742725166b06f",
+	token: "",
 	seed: SEED,
 	numDays: days,
 	numEvents: num_users * 120,
@@ -447,7 +447,10 @@ const config = {
 				record.sequence[0] === "checkout started" &&
 				record.sequence[1] === "order placed") {
 				record.props = record.props || {};
-				if (meta && meta.userIsBornInDataset) {
+				// Use hash to deterministically assign ~50% of users as "new"
+				const userId = meta && meta.user && (meta.user.distinct_id || String(meta.user));
+				const isNewUser = userId && userId.charCodeAt(0) % 2 === 0;
+				if (isNewUser) {
 					record.conversionRate = 0.90;
 					record.props.first_order_bonus = true;
 				} else {
@@ -941,7 +944,7 @@ export default config;
  *   import config from './dungeons/harness-food.js';
  *   const results = await generate(config);
  *
- * OUTPUT FILES (with writeToDisk: true, format: "json", gzip: true):
+ * OUTPUT FILES (with writeToDisk: false, format: "json", gzip: true):
  *
  *   - needle-haystack-food__events.json.gz - All event data
  *   - needle-haystack-food__user_profiles.json.gz - User profiles