Price Estimation Server Using Local Classification Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for providing price estimates for variable goods and services, such as airfares, are inefficient due to the need for frequent access to remote servers and extensive data storage, which is costly and impractical, especially when dealing with vast numbers of possible queries and routes.
Innovation Solution
A method that uses historical price datasets to analyze patterns and calculate estimated prices using classifiers, such as Naïve Bayes models, to predict prices without requiring access to Global Distribution Systems, thereby reducing the need for extensive data storage and server queries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods access remote servers frequently to obtain accurate price information, then price accuracy is improved, but response time and energy consumption increase
Solution Approach 1:
The system pre-processes historical price data and trains classification models in advance, storing the learned patterns locally. When a price query arrives, the pre-trained model immediately classifies the request and generates an estimate without needing to access remote servers, thus maintaining accuracy while dramatically reducing response time.
Solution Approach 2:
Instead of accessing the original price database on remote servers for each query, the system creates local copies of price patterns through trained classification models. These model copies capture the essential pricing logic and can independently generate estimates, eliminating the need for repeated server access while preserving pricing accuracy.
2Measurement precision
If extensive historical price data is stored for all possible routes and time horizons, then estimation accuracy is improved, but data storage requirements increase
Solution Approach 1:
The system transforms raw historical price data into classified patterns using training data with specific parameters (routes, time periods, price ranges). By changing the data representation from exhaustive storage to compressed classification models, the system maintains estimation accuracy across all possible queries while storing only the essential pricing patterns.
Solution Approach 2:
The system extracts the essential pricing patterns and relationships from extensive historical data during the training phase, separating the core pricing logic from the raw data volume. This extraction creates compact classification models that capture pricing behavior without requiring storage of all original historical records.
3Reliability
If remote server access is used to calculate fare prices, then price reliability is improved, but energy consumption and cost increase
Solution Approach 1:
The system makes the price estimation service self-sufficient by locally storing trained classification models that can independently process queries. The local model serves itself by generating price estimates without external server assistance, maintaining reliability through preserved pricing logic while eliminating the energy costs of repeated network access and remote computation.
Solution Approach 2:
All computational work for learning pricing patterns is performed in advance during model training, before actual price queries arrive. This preliminary computation embeds the pricing knowledge locally, so subsequent queries require minimal energy for simple classification operations rather than energy-intensive remote server access.
Data Source
AI summary
The field of the invention relates to methods, servers and computer program products for providing a set of prices. A computer server receives a request for a price for goods or services, such as airfares, together with parameters defining those goods or services, for example: activity type, such as airfare, hotel booking, train fare; date range; destination; origin; desired weather conditions; star ratings; keywords; any other user defined preference. One or more processors programmed with software then infer, estimate or predict estimated prices from an incomplete historical price dataset by analysing patterns in that dataset and provide the price estimates to an end-user computing device, such as a personal computer, smartphone or tablet.