N-Gram Sales Forecasting via Discretized Probability Tables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sales forecasting methods lack accuracy in evaluating multiple promotion plans for sales increase and profitability, particularly in handling integer sales values and scalability, with regression methods being preferred over classification due to superior performance and computational efficiency but failing to provide precise predictions for complex scenarios.
Innovation Solution
A system utilizing an N-Gram model for sales forecasting, which discretizes sales history into groups based on sales velocity and bin width, generates a probability table through weighted mean and median computation, and applies hybrid smoothing techniques to produce a smooth probability table for multistep predictions using joint, bootstrapped, or step greedy approaches.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If regression methods are used for sales forecasting, then prediction accuracy and computational efficiency are improved, but the ability to handle integer sales values and provide precise predictions for complex promotion scenarios deteriorates
Solution Approach 1:
The patent segments the continuous sales forecasting problem into discrete integer value predictions by introducing a probability distribution function that models sales as discrete units. The sales forecast is divided into multiple possible integer outcomes with associated probabilities, allowing the model to handle the discrete nature of sales data while maintaining the computational advantages of continuous probability modeling.
2Adaptability or versatility
If classification methods are used for sales forecasting, then integer sales value handling is improved, but prediction accuracy and computational efficiency deteriorate
Solution Approach 1:
The patent replaces traditional classification methods with a probabilistic regression approach that uses a probability distribution function to model discrete sales outcomes. Instead of assigning sales to discrete classes through classification, the system uses continuous probability density functions to predict the likelihood of different integer sales values, combining the strengths of both classification and regression approaches.
3Device complexity
If traditional forecasting models are used, then simplicity is maintained, but the ability to evaluate multiple promotion plans and adapt to complex scenarios deteriorates
Solution Approach 1:
The patent introduces dynamic adaptability by making the probability distribution function parameters adjustable based on different promotion scenarios. The model can dynamically adapt to various promotion plans by modifying the underlying probability distribution parameters, allowing evaluation of multiple promotion strategies while maintaining a relatively simple core modeling framework.
Solution Approach 2:
The patent enables evaluation of multiple promotion plans by allowing changes in the parameters of the probability distribution function based on different promotion conditions. By adjusting parameters such as mean, variance, or distribution shape according to promotion characteristics, the model can evaluate different promotion scenarios without requiring completely different model structures.
Data Source
AI summary
This disclosure relates generally to method and system for forecasting sales based on N-Gram model. The present disclosure provides accurate prediction of sales for optimal operations to reduce the cost. The method receives a plurality of inputs of each product comprising a sales history, and a current price bin. The categorical sale(s) for each product is discretized based on the sales history by clustering each product sales history into a one or more groups based on a maximum sales velocity range. Further, a probability table is generated for the discretized categorical sales of each product based on computing a round off weighted mean and a median using a N-Gram model. Then, a smooth probability table is computed for the generated probability table. To forecast sales multistep prediction for the smooth probability table is computed based on at least one of a joint approach, a bootstrapped approach, and a step greedy approach.