Business intelligence / governed data / dashboards with a pulse

Make the number
explain itself.

A full-stack BI exhibit: model the meaning, secure the platform, write the query, build the report, test the result, and make the decision obvious.

The business dashboard below uses a fictional company and entirely synthetic data. Separate global-signal exhibits use named public sources with visible timestamps and limitations. No employer, customer, visitor, portfolio, or personal records enter this lab.

Interactive dashboard / fictional Northstar Supply Co.

Executive pulse,
without mystery math.

100% synthetic dataset

No visitor tracking. No personal records. No employer or customer information.

Loading the governed query…

Dashboard filtersWindowRegionChannelSegment

Revenue

$16.4M+25.4% vs prior period

Orders

144,050+24.2% vs prior period

Margin rate

25.1%0.0 pts vs prior period

Customers

100,429+24.3% vs prior period

Trend explorer

Revenue by month

Contribution

Revenue by region

West$4.5M
27.6% share · 27.0% margin
Midwest$4.2M
25.9% share · 25.6% margin
Southeast$4M
24.1% share · 24.3% margin
Northeast$3.7M
22.4% share · 22.9% margin

Detail table

Latest governed result set

432 fact rows in the selected window

Latest synthetic analytics result rows
Period	Region	Channel	Revenue	Orders	Margin	Vs target
Jun '26	West	Marketplace	$160.1K	1,425	28.2%	-3.5%
Jun '26	West	Partner	$153.2K	1,356	27.3%	-3.9%
Jun '26	Midwest	Marketplace	$150K	1,324	26.8%	-3.9%
Jun '26	West	Direct	$146.2K	1,288	26.4%	-3.4%
Jun '26	Southeast	Marketplace	$143.1K	1,250	25.4%	-3.4%
Jun '26	Midwest	Partner	$143K	1,256	25.9%	-3.4%
Jun '26	Midwest	Direct	$139.3K	1,215	25.0%	-2.8%
Jun '26	Southeast	Partner	$136.1K	1,182	24.5%	-2.8%
Jun '26	Northeast	Marketplace	$129.8K	1,122	24.0%	-2.8%
Jun '26	Southeast	Direct	$129.1K	1,114	23.6%	-3.2%
Jun '26	Northeast	Partner	$126K	1,081	23.1%	-3.2%
Jun '26	Northeast	Direct	$122.3K	1,040	22.2%	-3.5%

Open the SQL query inspector

The public API accepts only published filter values, binds each value as a parameter, caps the result, and exposes no write method. The dashboard receives aggregates; it does not receive a database console.

SELECT period_start, region, channel, segment,
       revenue_cents, target_revenue_cents,
       margin_cents, orders, customers, returned_orders
FROM analytics_facts
WHERE period_start BETWEEN ? AND ?
  AND (? = 'all' OR region = ?)
  AND (? = 'all' OR channel = ?)
  AND (? = 'all' OR segment = ?)
ORDER BY period_start, id
LIMIT 1000;

Cognos administration / capacity thought experiment

Keep the service calm
when the schedules wake up.

Explore a simplified dispatcher model. Real sizing also depends on report complexity, data-source latency, memory, process settings, affinity, burst behavior, and measured workload. This is an operations lesson, not a substitute for performance testing.

Dispatchers 3Report processes / dispatcher 4Peak concurrent requests 18Reserve one dispatcher for failover

Illustrative load113%

Queue pressure

Rebalance schedules, inspect workload, and test capacity before adding traffic.

Active dispatchers: 2
Teaching capacity: 16 slots
Failover reserve: 1 dispatcher

Python forecast workbench / deterministic sandbox

Change the assumption.
Watch the forecast answer.

The browser preview mirrors a compact Python transformation: establish a baseline, apply a controlled trend, add a seasonal wave, and expose every assumption. It is an explanatory demo, not a production forecast.

Annual growth assumption 8%Seasonality amplitude 6%

Six-month scenario$10.3M

$1.8MM1

$1.8MM2

$1.7MM3

$1.6MM4

$1.6MM5

$1.7MM6

from math import sin, pi

baseline = 1678230
annual_growth = 0.08
seasonality = 0.06

forecast = [
    round(
        baseline * (1 + annual_growth * month / 12)
        * (1 + sin(month / 6 * 2 * pi) * seasonality)
    )
    for month in range(1, 7)
]

Global systems monitor / public sources / provenance first

Watch the world.
Keep the labels honest.

One map canvas joins a static strategic-reference sample with calculated civil satellite positions and selected-region civilian aircraft. Every mode keeps source status, limitations, and a keyboard-first ledger in view.

Strategic sampleSnapshot24 rounded public references / non-operational

Orbital watchConnecting0 positions / Source time unavailable

Flight watchConnecting0 open-transponder records / Source time unavailable

Modeled motion / public orbital elements / 1-second updatesCivil & science satellites

Hundreds of current positions are calculated in this browser every second from commercial constellations and an explicit civil spacecraft allowlist using satellite.js SGP4. They are modeled estimates, not direct spacecraft telemetry; broad topical groups are filtered by catalog ID.

ConnectingAwaiting sourceSource time unavailable

Visible orbital watch layers

Global Mercator fallback0 visible map records

Opening world view

Opening the global map.

ConnectingConnecting to the same-origin data service.

CrewedScienceWeatherEarth observationNavigationCommunicationsOther civil

Markets + economy / source-aware live desk

Read the tape.
Respect the clock.

Live exchange candles move by the second. Economic facts move when an agency publishes them. This desk keeps those clocks separate, shows the source, and never turns an indicator into a recommendation.

Day Trader Workbench

Connecting to Kraken public market data…

BTC / USD—Awaiting current candle

Waiting for market candles

Session VWAP—Volume-weighted reference

EMA 20—Short trend estimate

RSI 14—Momentum, not a signal

ATR 14—Average candle range

Realized vol.—24h 5-minute log returns

Spread—Top-of-book snapshot

Book imbalance—Top 25 levels

24h range—Observed candle range

U.S. Economic Pulse

Loading the latest official releases…

Labor—Awaiting official release

Prices—Awaiting official release

Payrolls—Awaiting official release

Federal debt—Awaiting official release

Release historyUnemployment rate

Waiting for the latest official release series.

Platform fluency / the whole machine

Six disciplines.
One governed answer.

Cognos, QuickSight, and Superset are distinct platforms, not one comma-separated feature. Each has its own architecture, governance model, deployment patterns, strengths, and operational vocabulary.

Enterprise platform

Cognos Administration

Namespaces, capabilities, schedules, dispatchers, content-store hygiene, deployment packages, auditing, capacity planning, and the calm art of finding the one setting everyone swears they did not change.

Security
Scheduling
Deployments
Operations

Authoring studio

Cognos Report Development

Prompted reports, lists, crosstabs, charts, drill-through paths, bursting, conditional layouts, reusable queries, performance tuning, and outputs that answer the business question before decorating it.

Reports
Prompts
Drill-through
Performance

Semantic engineering

Framework Manager

Import, business, and presentation layers; star schemas; determinants; cardinality; reusable calculations; dimensional behavior; governors; and names that make sense to someone who did not build the warehouse.

Metadata
Cardinality
Determinants
Governance

AWS-native analytics

Amazon QuickSight

SPICE and direct-query tradeoffs, datasets, row-level security, calculated fields, parameters, controls, scheduled refreshes, and focused dashboards designed for cloud distribution.

SPICE
RLS
Calculated fields
Cloud BI

Open-source BI

Apache Superset

Semantic datasets, virtual metrics, SQL Lab, native filters, role-based access, caching, dashboard composition, and the practical responsibilities that arrive with an extensible platform.

SQL Lab
Datasets
RBAC
Caching

Shared foundations

SQL, Python & BI Design

Dimensional modeling, CTEs, window functions, reconciliation, data quality, profiling, automation, forecasting, requirements, lineage, and the final mile: explaining what the number means without performing interpretive dance.

SQL
Python
Modeling
Data quality

Framework Manager / semantic layer anatomy

The report is only as trustworthy as the model beneath it.

A durable BI system separates storage, business meaning, and presentation. That separation lets one definition of revenue serve a board report, an operational dashboard, and an analyst without becoming three revenues.

01
Source layer
Typed facts, dimensions, dates, keys, and quality rules. The warehouse remains gloriously unimpressed by font choices.
02
Business layer
Joins, cardinality, determinants, conformed dimensions, reusable measures, and definitions that survive beyond one dashboard.
03
Presentation layer
Friendly names, subject areas, curated measures, navigation paths, security, and just enough choice for confident self-service.
04
Consumption
Cognos reports, QuickSight analyses, Superset dashboards, governed exports, alerts, and decisions with an audit trail.

Architecture choices / no tool cosplay

Pick the operating model, not merely the logo.

High-level platform comparison for the fictional analytics lab
Platform	Natural strength	Modeling center	Operational focus
IBM Cognos Analytics	Governed enterprise reporting	Packages, modules, Framework Manager	Administration, scheduling, security, bursting
Amazon QuickSight	AWS-native distribution and embedding	Datasets, calculated fields, parameters	SPICE, refreshes, RLS, cloud consumption
Apache Superset	Open, extensible exploration	Datasets, virtual metrics, SQL Lab	Deployment, RBAC, caching, observability

The quality gate / before anyone says “the dashboard is wrong”

Reconcile. Profile. Document. Then publish.

01
Confirm grain, keys, date logic, and join cardinality.
02
Reconcile totals to a controlled source at every transformation boundary.
03
Test security, prompts, empty states, exports, schedules, and edge cases.
04
Record definitions, owners, lineage, refresh timing, and known limitations.