El proceso Ágil Kanban

An influence diagram approach to automating lead time estimation in Agile Kanban project management

Juan Sebastian Godoy Hernandez a,1, Sebastian Parra Losada a,2, Andres Felipe Tenjo a,3, Lizbeth Yuliza Cepeda Sarmientoa,4

a Specialization in project management in engineering, University of La Salle, Race 5 No. 59A-44. Col
Juan Sebastian Godoy jgodoy77@unisalle.edu.co

Sebastian Parra sparra66@unisalle.edu.co

Andrés Felipe Tenjo antenjo05@unisalle.edu.co

Lizbeth Yuliza Cepeda Sarmiento lcepeda25@unisalle.edu.co

Keywords: Kanban, Estimate, System, Project, Team, Expert, Model

Palabras claves: Kanban, Estimación, Sistema, Proyecto, Equipo, Experto, Modelo

Abstract— This research presents an approach based on influence diagrams to estimate task delivery times for Agile Kanban project management. While these factors can be critical in Scrum, teams practicing Kanban frequently experience these conditions due to the emphasis on continuity, integration, and reprioritization. To partially automate the estimation process, an expert system based on influence diagrams is developed and populated with data from a practicing Kanban team and used to generate a cumulative distribution function to facilitate communication of probabilistic estimates.

Síntesis— Esta investigación presenta un enfoque basado en diagramas de influencia para estimar los tiempos de entrega de tareas para Agile Kanban gestión de proyectos. Si bien estos factores pueden ser críticos En Scrum, los equipos que practican Kanban experimentan estas condiciones con frecuencia debido al énfasis en la continuidad integración y Re-priorización. Para automatizar parcialmente el proceso de estimación, un Se desarrolla un sistema experto basado en diagramas de influencia y se completa con datos de un equipo Kanban practicante y se utiliza para generar una función de distribución acumulativa para facilitar la comunicación de estimaciones probabilísticas.

1. Introduction

T This research presents the application of an expert system based on influence diagrams to represent the uncertainty in the estimation of the delivery time for the management of Agile Kanban projects. Project managers in the software industry or other fields prone to rapid change or high levels of uncertainty have largely embraced agile project management methodologies, and while most teams practice Scrum, the newer process Kanban has been growing rapidly. A hybrid of Scrum and Kanban called Scrumban has also been gaining traction recently. This usually happens during a multi-hour planning meeting in Scrum or during the daily meeting in Kanban.

When story point estimation is applied to Agile Kanban, which can be thought of as a lightweight version of Scrum that skips meetings out of daily standup, the level of overhead required to support the estimation becomes even more disproportionate to with other administrative tasks of the team. However, Kanban by design does not provide clear milestone dates for delivery like those provided by detailed upfront planning and Gantt charts used by traditional waterfall project management. Instead, Kanban pulls the task through a workflow pipeline as resources become available. This absence of milestones can lead to a perceived lack of predictability, which can increase reluctance within an organization to transition teams to agile methodologies, including Kanban.

When planning poker for story point estimation is applied to Kanban projects, the incorrect assumption is made that the project backlog will remain unchanged. In reality, Kanban seamlessly integrates tasks, allowing them to be added to or removed from the backlog, or reprioritized at any time . Due to these challenges, more advanced analytical models must be developed to improve the estimation capabilities of Kanban teams, reduce time spent on non-value-added activities, improve information communicated to stakeholders, and aid in the decision-making process. task prioritization. We developed our influence diagram through literature review and input from a professional Agile Kanban project team.

This expert system was populated with data collected from a professional Kanban project team at our industry partner Andersen Corporation, a leading North American manufacturer of window and door products, as a demonstration of how the resulting information can be implemented to reduce the time team members spend producing lead time estimates and how it can be used as a communication tool for stakeholders. This research designs an expert system based on a probabilistic model of task movement within the Kanban project backlog with the aim of reducing the input required by the team and accounting for the factors that influence task movement, which are ignored by the team. current estimation techniques. In Section 2, we present an overview of Kanban that was used to develop the structure of the influence diagram and a general mathematical description of influence diagrams. Section 3 covers creating the graphical influence diagram model, collecting data from the Kanban team, and creating the interactive tool in Netica.

Finally, Section 5 summarizes the conclusions obtained during the development of this estimation tool and looks forward to the next steps to further automate Agile Kanban project management and other influence diagram applications in expert systems.

1. INTRODUCCIÓN

Esta investigación presenta la aplicación de un sistema experto basado en diagramas de influencia para representar la incertidumbre en la estimación del tiempo de entrega para la gestión de proyectos Agile Kanban. Los gerentes de proyecto en la industria del software u otros campos propensos a cambios rápidos o altos niveles de incertidumbre han adoptado en gran medida metodologías ágiles de gestión de proyectos, y aunque la mayoría de los equipos practican Scrum, el proceso más nuevo Kanban ha estado creciendo rápidamente. Un híbrido de Scrum y Kanban llamado Scrumban también ha estado ganando terreno recientemente. Esto suele ocurrir durante una reunión de planificación de varias horas en Scrum o durante la reunión diaria en Kanban.

Cuando la estimación del punto de la historia se aplica a Agile Kanban, que se puede considerar como una versión ligera de Scrum que salta las reuniones fuera del standup diario, el nivel de gastos generales necesarios para respaldar la estimación se vuelve aún más desproporcionado en relación con otras tareas administrativas del equipo. Sin embargo, Kanban por diseño no proporciona fechas claras de hitos para la entrega como las proporcionadas por la planificación inicial detallada y los diagramas de Gantt utilizados por la gestión de proyectos en cascada tradicional. En cambio, Kanban extrae la tarea a través de una canalización de flujo de trabajo a medida que los recursos están disponibles. Esta ausencia de hitos puede conducir a una percepción de falta de previsibilidad, lo que puede aumentar la renuencia dentro de una organización a hacer la transición de los equipos a metodologías ágiles, incluido Kanban.

...