Overview of Learning Models
A Learning Model is an algorithm that can learn from and make predictions on data gathered during the booking process.
The IBE uses this data to identify the best possible configuration for a property.
The IBE identifies the best case scenarios, and then optimizes the booking experience by adjusting the configurations in real time.
There is an unlimited amount of optimizations to choose from.
The Learning Models guarantee a strategy that strives for better outcomes for IBE users.
How to see what decisions have been made
Follow the steps below to navigate to the Learn Model results
Go to the AI module.
Go to the Overview page if you want to see a brief snapshot of the results to date.
Go to the Learning Summary page if you want to see deeper detailed results.
Learning Model decisions overview
Models are ordered with those that have definitive results for your property being ranked higher, followed by most recently examined models.
Where a winning model path has been identified, this is denoted with an icon overlaying the winning path.
Once a decision has been made, the booking engine will automatically reconfigure itself to run with the behavior defined in that model path.
A summary of the dates on which the model was examined for your property is also visible.
Clicking on the model title to open a popup which will explain more about the configuration defined in the model.
Learning Model Results
A set of results for each model is displayed.
For each model, two horizontal rows are shown
The top row shows the number of unique visitors that have been presented with the respective model path.
The bottom row shows the number of unique visitors that have reached the model goal for each path. From these two figures, the IBE can determine the effectiveness of each path. Where a statistically significant model path is identified, a visual icon is overlayed on the path.
The conversion rate and conversion rate range, which allows for standard error considering sample size, is displayed.
The IBE will determine a model result when there is a clear winning path and at least 3000 visitors have been sampled across all model paths.
Where there is insufficient data to determine a winning path, the model will continue to be examined until such time as sufficient data exists to make a determination.
Smaller visitor numbers result in a wider error margin. The IBE will not make any adjustments until there is more data.
Learning Models Properties are configured for
The Overview section under My Learning Models holds a list of models that are enabled for your property.
You can see how many unique visitors have been presented with each model path, and how many of those visitors have achieved the model goal, which gives the success rate of each model.
To change the order in which models are examined, re-order models by dragging them into a higher position.
Any models that are ordered above the dotted grey line are your priority models and will be examined in the order you determine.
Note: You can add a maximum of 5 priority models.
All other models will be examined in an order determined by the IBE.