Product platform development

Companies are valued by their market competitiveness. To be competitive in the market, companies have to respond to their customer requirements as best as possible. To this end, companies introduce new products to the market to better capture customer requirements. The use and employment of market data have been widely studied in academia and industry. The trend in customer can be determined from market data and can bring useful and interesting knowledge of the true customer requirements. In this thesis, a systematic methodology of "Product Platform development" is proposed which uses market sales data. The proposed methodology is composed of several steps. The steps each take advantage of one or several algorithms. The theoretical background of each step of the methodology and the algorithms are presented in the thesis. The first step of the proposed methodology is to normalize the input data and perform robust fuzzy clustering of normalized data. The second step is the extraction of both positive and negative association rules. The positive association rule concept is employed to find the relationships between customer requirements and functional requirements. Some of the general early stage design constraints can be identified by using the negative association rules concept is used to identify which can facilitate the product concept generation and selection procedure. An algorithm is developed to find the positive and negative association rules and to select the correct set of association rules. The extracted association rules are developed in a way which makes it possible to have minimum length antecedents and maximum length consequents. To this end, some modified algorithms have been proposed which not only consider the job of finding frequent itemsets from the fuzzy clusters of functional requirements but also extract the non-redundant association rules. During the third step, trends in customer requirements are identified by tracking the changes of association rules over several consecutive time periods. These trends are identified by extracting meta-rules. Finally, a new algorithm to aggregate the extracted association rules helps the product development engineers in making decisions about the desirability of possible product platforms. Given the different extracted association rules and the meta-rules, rules that show the trend of changes in the customer requirements and the products that are bought, the possible future product platforms are proposed as the stable part of a product family which can meet customer requirements. Modular software to investigate the performance of methodology was designed which allows the user to accomplish the whole proposed methodology without the need for any other software. This software allows future methodology extensions to be easily implemented.