Project Scheduling

Project Scheduling
An Introduction

Prof. David Bernstein
James Madison University

Computer Science Department

bernstdh@jmu.edu

Motivation

The Question:
- How long will it take to complete a specific (design or implementation) project (or part of a project)?
A Naive Approach:
- Add the effort estimates for each task
An Overview of the Shortcomings of the Naive Approach:
- Units Matter
- Task Details Matter
- Task Dependencies Matter
- Personnel Capabilities Matter

Units Matter

Effort (\(E\)):
- Measured in person-days or person-months
Time (\(T\)):
- Measured in days or months
The Implication:
- Thinking only about units, \(T = E / N\), which means that \(N\) (i.e., the number of people working on the project) matters

Task Details Matter

A Property of Very Small Tasks:
- They can only be worked on by one person at a time (hence, \(T = E / N\) only applies when \(N = 1\))
A Property of Many Tasks:
- Large values of \(N\) can make things worse (the famous "Mythical Man-Month")

Task Dependencies Matter

Personnel Capabilities Matter

An Observation:
- Different members of the team have different knowledge, skills, and abilities
Implications:
- Personnel may make it impossible to complete tasks in parallel
- A "person" is not a homogeneous thing

An Example You Have Experience With

An Example You Have Experience With (cont.)

Another Question You Might Ask:
- Which courses can't I afford to miss or fail?
Generalizing This Question:
- Which tasks can/can't be delayed without delaying the completion of the project?

An Example

The Setting:
- The design and implementation of a new computer game
Where we are in the Process:
- The major tasks in the process have been identified at a low level of detail
- The time required for each task has been estimated
- The task dependencies have been identified

An Example (cont.)

The Time Required for Each Task

An Example (cont.)

The Task Dependencies

Solving the Problem

Some Observations:
- This problem is not unique to software projects
- There are many ways to solve it (and variants of it)
Techniques: