Forecasting Inter-generation Product Transitions

Project: Research project

Project Details

Description

Forecasting Inter-generation Product Transitions Forecasting Inter-generation Product Transitions The objective of this research is to develop a systematic method of demand forecasting for product transitions at Intel. Deliverables will include demonstration of the forecasting method on selected Intel products and research papers and seminars. The investigators will hold regular meetings with Shamin Shirodkar, Karl Kempf and other Intel personnel, as well as offering guidance for Shamin's research project and thesis. The investigators of this research project include Hongmin Li, assistant professor in Supply Chain Management, and Dieter Armbruster, professor in Mathematics, both from Arizona State University. The research team will also include a full time student research assistant, from either the Business school or Math department, for the duration of the project. We will start by building and testing the Logit model for forecasting at the product family level and at the monthly time granularity. This involves analyzing existing Design Wins and PTI data and identifying key product characteristics that influences demand. A combined data set for three product families has been pulled from both the Design Wins and PTI database for that purpose. In the first stage of the research (a period of 12 months), the investigators and the Intel product transition team will work together to rationalize and test the basic aggregate model (without differentiating customers into specific types). In the second stage (a period of 12 months), we will extend the model to handle more aspects of competition, and build an agent based simulation model to simulate the behavior of our customers. In the third stage (year three), we explore the develop diffusion models or their combinations for improved forecasting, and compare the performances and applicability of these models against the choice models. The risks to the project are the availability of adequate data to quantity a large number of model parameters (as shown in equation (1)). We plan to ease the problem as needed by discussing with the Intel managers and engineers to reduce the number of parameters to focus on the most critical factors to get around this issue. Also we will put additional effort to provide a solution that has high explainability of results to make it easy to understand and applicable for business use. We plan to work very closely with our customers internal to Intel to ensure we capture the detailed business problem and will develop User Interfaces which support the explainability to increase the probability of delivering a high useful solution.
StatusFinished
Effective start/end date10/1/086/30/13

Funding

  • INDUSTRY: Domestic Company: $80,000.00

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.