Hotel data, turned into decisions

Working examples of how a multi-property hotel group can use the data it already has to sell more rooms at the right price, schedule staff fairly, and remove avoidable cost. Everything here runs on one shared sample dataset of five fictional hotels.

By Nihal Python, SQL, optimization, HTML dashboards All hotels, guests and staff are fictional
Start here

Working demos

Open either page. No login or installation is needed.

Revenue Command Centre

Three years of booking history turned into a plan for every night of the next twelve months: which nights will sell out, which need help, and what to do about each.

Open the Revenue Command Centre

Staff Schedule Builder

A live tool anyone can use: enter the week's workload and your team, press Build, and get a schedule that checks every labour rule, honours day-off requests and exports to Excel or print. Your inputs are saved as you type. The figures below are for the sample week, using the full optimisation model.

Open the Staff Schedule Builder
New

Free hotel tools

Everyday planning jobs that hotels usually pay for, as free and simple tools. Type your own numbers over the sample and get an answer in plain English. Nothing is sent anywhere.

See all tools
💲

Rate Recommender

A suggested rate for each of the next 30 days, with the reason.

 🛏️

Overbooking Calculator

How many extra rooms to sell so no-shows don't leave rooms empty.

 🌙

Night Audit Assistant

Checklist, balancing, rate checks and the daily manager's report.

 🧹

Housekeeping Board

Today's room assignments by who arrives first, with printable boards.

 📈

Occupancy Forecaster

The next 30 to 90 days from your own history, with a likely range.

 👥

Group Quote and Displacement

Is this group worth more than the guests it pushes out?

 📊

Hotel KPI Dashboard

Occupancy, ADR, RevPAR and profit from a spreadsheet upload.

 ⏱️

Labour Productivity

Hours, FTEs and cost per occupied room at every occupancy level.
AI research

Hotel Swarm Operations

AI agents that coordinate peak-hour energy, housekeeping, overbooked guests and open shifts across the five sample hotels, measured against how hotels handle them today. In the sample: $279k a year in peak-energy savings and $14.8k better per oversold night.

See the results
The ideas

Proposals

Each proposal explains what the idea is, why it matters with linked sources, and how it would be solved and measured. Project HARLTON covers revenue, profit and expense; Project H2 covers operations such as staffing, guest flow and facilities.

Project HARLTON: revenue, profit and expense

Project H2: operations

Shared foundation

For technical reviewers

How it is built

One pipeline produces every number on this site. It can be rebuilt from scratch with one command: python analytics/build.py

Sample dataBooking requests replayed against room inventory, so sell-outs, denials and cancellations behave realistically.
SQL warehouseStar schema: hotel, date, segment and channel dimensions; booking, stay-night and denied-night facts.
Quality checksTen checks, including revenue tie-out between grains, must pass before anything is published.
AnalyticsBooking pace, pickup forecast with back-test, unconstrained demand, demand tiers, schedule optimization.
This siteStatic pages reading exported JSON, plus Excel downloads with live formulas.

Code: analytics package · SQL scripts · tests · sample data extract

About this portfolio

I built this to show how I would approach business intelligence in hospitality: start from the decisions people make every day, use the data the hotel already collects, check that every number ties back to its source, and present it so anyone can act on it.

The proposals, analysis approach and direction are my own. AI tools were used for research, drafting and coding support; they are credited in the repository README. No employer data is used anywhere.