Demand Prediction by Customer Segmentation for Booking Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for estimating service bookings do not accurately account for the varying booking tendencies of customers based on their attributes, leading to inaccuracies in predicting demand.
Innovation Solution
Derive separate exponential functions for different customer groups based on their attributes, such as loyalty or price sensitivity, to improve the accuracy of demand prediction and price determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single exponential function is used to estimate bookings for all customers, then the model is simple and easy to implement, but the prediction accuracy deteriorates because it cannot account for varying customer attributes
Solution Approach 1:
The patent segments the customer base into multiple customer groups based on attributes such as loyalty status and price sensitivity. Each customer group is assigned a separate exponential function with its own slope parameter, allowing the model to capture different booking tendencies for different segments while maintaining the overall simplicity of the exponential function form.
Solution Approach 2:
The patent applies local quality by allowing different slope parameters in the exponential function for different customer groups. This means that while the overall model structure remains consistent, the specific parameters are locally optimized for each customer segment, improving prediction accuracy without requiring a completely different model for each group.
2Measurement precision
If separate exponential functions are derived for different customer groups, then the prediction accuracy improves, but the computational complexity and data processing requirements increase
Solution Approach 1:
The patent performs preliminary action by pre-segmenting customers into groups based on their attributes before the booking estimation process. This segmentation is done once and stored, so that during actual prediction, the system only needs to identify which pre-defined group a customer belongs to and apply the corresponding exponential function, rather than performing complex real-time analysis.
Solution Approach 2:
The patent uses parameter changes by adjusting the slope parameter of the exponential function based on customer group attributes. Instead of using a single fixed slope for all customers, the system changes the slope parameter according to the customer's loyalty status and price sensitivity, allowing flexible adaptation to different customer behaviors while maintaining the same functional form.
Data Source
AI summary
A demand prediction device derives a first exponential function representing the time-series transition of the number of bookings until a service provision time point for a first customer group based on booking transitions up to a first time point t1. A second exponential function is similarly derived for a second customer group based on booking transitions up to a second time point t2 different from t1. The device generates information supporting a service provider based on the predicted time-series transitions for the first and second customer groups. A time constant is compared with at least one threshold value to determine a system response, which triggers an adjustment to one or more operational parameters, including modifying inventory levels, adjusting reservation or scheduling limits, reallocating staffing or system resources, or updating pricing or promotional parameters. This enables dynamic demand prediction and responsive operational optimization for service provision.