Determining SOW of Product Release by Cluster Analysis of Project Activities

The proposed research deals with determining the scope of work (SOW: project components, features as well as development activities) of releases in Continuing Technology Development projects (CTD projects), including hardware and/or software and/or R&D projects, by utilizing cluster analysis approach. The suggested method generates the SOW by associating the developed components to releases based on similarity between their features, such as common functionality, tasks, and activities. In addition, we propose a special type of distance measure based on competition for common resources as well as complementary or dissimilar nature.

The current methodology of project scheduling focuses mainly on the concept of a project as a one-time concentrated effort. However, CTD projects are the building blocks of a continuing activity, which is divided into several intermediate projects leading to product Releases. Each release is composed of several new features, and the development of each feature requires a set of activities (some may overlap with activities of other features). Current planning methodologies ignore the unique characteristics of the CTD projects. In particular, the division to releases is overlooked in current project planning methodologies

The CTD project scheduling goal is to include the features in as earlier release as possible (so as to present an attractive product to the market and beat the competitors), subject to limited resources. Thus, the scheduler has to determine the components of each release (i.e. features included in release). This decision is based on the expected benefit from the implementation of the new feature and on the efforts and the resources needed to implement all the required activity. This presents a complex problem as each project task is related to several features, and also competes on the same resources.

The current research proposes a new methodology for the planning problem, by clustering the components into releases. The similarity of two features will be based on the weighted sum of efforts, resources and benefits related to their overlapping as well as distinct activities.  Similarity measures will be developed to reflect the similar nature of the activities.

The research presents a novelty in two aspects:

• This is the first methodology to tackle the progress of a CTD project and planning timeline of releases. CTD planning divides the activity into product releases. This includes the determination of the new features as well as the scheduling and releases timeline. So far, this problem was tackled solely by intuition, lacking a proper decision making process. This research will provide a decision support tool to answer questions such as "what to include\exclude in the next release?" and "What benefit can be gained from postponing a release?"
• The adaptation of clustering methods to CTD planning and scheduling.