|M.Sc Student||Amit Yurman|
|Subject||Product Cost Estimation Using Grouping Techniques|
|Department||Department of Industrial Engineering and Management||Supervisor||Mr. Rubinovitz Jacob (Deceased)|
|Full Thesis text - in Hebrew|
The costing problem of new product in Engineer-To-Order organizations is acute. Doing the whole engineering work (and evaluating the product cost) for each sales quotation requires a lot of time and resources that do not justify themselves. Evaluating the cost of an identical item is not an option and the salesman is forced to find different methods to estimate the product cost. The problem is even more complex because 1) in most cases the salesman does not have all the engineering data at the time of the quotation, and 2) the need to know the cost variation as well.
This research examines methods to find the “nearest” products, based on the engineering data that is available. We examined two methods for grouping products: finding the K-nearest neighbors and filtering the database. The filter method uses iterations, in which the search interval gets wider.
The two main innovations in this research are: 1) using “probability distance” - which counts the number of records between two items - and not the common physical distance, and 2) developing the dynamic filtering method.
The quantitative analysis was based on real data and concludes that the use of the most innovative method -“probability distance” with the dynamic filtering - predicts the best results.