HS | AI Forecasting

AI Forecasting from Fourth is here! Using state-of-the-art machine learning algorithms to analyze your company’s data, the AI Forecast quickly detects patterns, uncovers trends, and identifies event outliers so you can confidently plan for future demand.

Please read the below FAQs to learn more about AI Forecasting in HotSchedules.

What is AI Forecasting, generally speaking?

AI forecasting is the use of artificial intelligence and machine learning to analyze historic data patterns in order to make predictions about the future without oversight.

What is Fourth’s AI Forecasting?

The Fourth AI Forecast Engine is powered by machine learning, a branch of artificial intelligence where data sits at the center of the solution. We are leveraging the latest developments and the use of algorithms that are powerful in automatically learning historical trends and patterns. These algorithms routinely outperform expert rules as well as simple time-series algorithms.

How does the AI Forecast look in HotSchedules?

In HotSchedules, the AI Forecast appears just like any other forecast you might be used to seeing — the difference is that it’s automatically generated by our AI Forecast Engine daily, which predicts your labor demand needs and allows you to Generate Labor.

You’ll know an AI Forecast has run by checking Forecast History in the Forecasting page (click the three-bar menu → Forecast History). In the history list, AI-generated forecasts will have a blank Author field since they aren’t entered by a user.

Once the forecast is in place, managers can review and, if needed, adjust it (if they have the appropriate permissions). The AI Forecast will only re-run automatically until either:

• An adjustment is made, or

• Labor is generated for at least one job for that week.

When an adjustment is applied, the Adjustment field will display details of the changes made to that forecasted week. If no adjustment is present, the field will display none.

If labor generation has occurred, you’ll also see the Labor Generation field update, as shown in the example below.

What is the Difference between Standard Forecasting and AI Forecasting?

There are 3 main differences to consider when deciding what forecasting model is right for you. 

1. Manager Input: Standard forecasting requires managers to select what weeks the model should analyze to predict demand. AI Forecasting is 100% automated requiring no manager input in selection of data. 
2. Demand Variability: If your business has stable demand, standard forecasting may be sufficient. However, if your demand fluctuates or you have heavy GM turnover, AI Forecasting detects event outliers from 7+ years of data so you can better predict demand and not be reliant on manager experience or tenure. 
3. # of locations/ROI: When moving from Excel to Standard forecasting, you can see your forecasting accuracy improve up to 75%. This is great for small operators looking to gain insights into demand while still having the capacity to review and adjust. As you begin to scale your operations, AI Forecasting allows you to create a personalized demand forecast for every location automatically. When moving from some form of advanced forecasting to Fourth’s AI Forecasting, our results show an average of a 20% improvement.

How far out does the AI Forecast predict?

The AI Forecast is currently configured to generate 42 days (6 weeks) into the future.

The AI Forecast is 100% automated, but can my manager still make adjustments?

Yes, managers can make adjustments to the AI Forecast. Some examples of this may be your manager preparing for a 30-top reservation or construction in your parking lot that will affect availability – things an AI Forecast could not know by analyzing past data/upcoming weather or seasonal events.

If you do NOT want managers making any adjustments, however, this can be disabled in HotSchedules.

How will we know when the forecast is locked?

If a manager makes an adjustment to the AI Forecast, the forecast for that week will automatically lock.

If no adjustment are applied and they are using our Labor Optimization, then it will lock when they generate labor for at least one job for the given week. Managers will know it's locked because the forecast for that week will show the timestamp of its last update e.g. Generated on Tue, Jun 17, 2:40 PM.

How does the AI Forecast algorithm work?

Fourth’s AI Forecast delivers unmatched accuracy. The algorithms follow a 3-step process to create your demand forecast.

1. Data Collection: Fourth’s AI Forecasting algorithm performs an in-depth analysis of intrinsic and external data points including 7+ years of historical sales data, guest counts, in-store prom
2. Data Pre-processing: The algorithm cleans the data by detecting and removing event outliers, increasing the accuracy of the forecast.
3. Data Training: The algorithm trains itself by analyzing its forecasted vs. actual results. With each new data point, the algorithm continues to refine the equation, making daily improvements.

How do you measure the accuracy of AI Forecasting?

We use weighted absolute percentage error (WAPE) to measure forecasting accuracy. WAPE takes into account the accumulation of error through differences between forecasted and actual values, therefore not masking this in the same was a variance over a period would have historically.

Does the AI Forecast tell you how much labor you need?

No, neither our standard nor AI Forecast tell you how much labor you need. A demand forecast predicts the volume types enabled for a customer, e.g., revenue, guests, products, etc. required to support planning of labor around these drivers.

We highly recommend customers use the AI Forecast in conjunction with HotSchedules Labor Optimization tool. A forecast can only drive promised results when executed correctly. With Labor Optimization, your managers can automatically create schedules that meet the forecasted demand AND your labor rules, so they are always properly staffed and you’re operating at peak efficiency. This is called demand-driven scheduling. 

Why did Fourth Invest in AI Forecasting?

HotSchedules first released Standard Forecasting several years ago with the same goal in mind as AI Forecasting – providing data-driven insight to enable better in-store planning. Advances in the field of machine learning over the past few years have skyrocketed, enabling restaurants to leverage the thousands of data points they accumulate every shift to make better-informed decisions moving forward.

When deciding what new technology to invest in at Fourth, our top priority is to improve the lives of the operator, removing non-value add work from their plate. AI Forecasting will not only improve your forecasting accuracy, helping your managers make more informed decisions around scheduling, but completely removes the burden of forecasting from their to-do list when creating schedules.