More Accurate Forecasts
The situation
The client needed a more accurate way of carrying out their revenue and occupancy forecasting. They were relying on a statistical technique called ARIMA (Autoregressive Integrated Moving Average), which wasn’t able to consider special events such as property closures.
The opportunity
To include special events and other, non-seasonal factors for more accurate revenue and occupancy forecasts.
How we made it happen
We worked closely with the client to introduce a new forecasting method using Prophet, an open-source machine learning tool. Prophet allowed revenue and occupancy forecasts to account for special events such as property closures, alongside normal patterns of demand.
What changed
- Revenue and occupancy forecasts became more accurate and more useful
- Special events were factored into forecasts rather than treated as anomalies
- Planning and decision-making improved across the business
- The impact of events and closures on client bottom line became clearer
Why it matters
Planning improved because the business could see the financial impact of special events and closures in advance, not after the fact.
