485 / 2024-04-25 16:30:50

Waterfall vs. Agile Project Choice

project management; Waterfall, Agile, Hybrid; iterative learning process, linear discriminant analysis

Project management supports approximately 30% of the world’s economy, with an annual value of $27 trillion. Traditionally, project management uses an intensively planned, sequential process known as Waterfall. However, for applications with uncertain requirements, an iterative learning process known as Agile has recently become widely used as an alternative. Still, for a broad range of high value applications, for example new product and service development, both processes are competitive. Project companies typically address the choice of project management process by using scoring models based on subjective weighted evaluation of relevant factors. As a more precise and reliable alternative, our work describes optimization models that incorporate random progress during project execution to support this choice. The Agile process is modeled using learning over multiple iterations, with customer feedback to guide product redesign. This model evaluates a measure of total expected total cost, including development cost, time-to-market, and closeness in design to the uncertain target market. A similar model of the Waterfall process receives customer feedback only at the end. The comparison of expected total cost recommends a choice between the two processes. Extensive sensitivity analyses provide insights about the robustness of this choice. We also develop a simpler procedure, with an expected loss of only 1.1%, for recommending a choice of process. This procedure is applied to two Hybrid models that combine Waterfall and Agile processes, to determine optimal switching times between them. We analyze a project from a leading game manufacturer and justify a recommendation for an Agile process which the company used with success.