BACKGROUND
A business unit of a company develops products under initial planning, which is sometimes strongly influenced by the appearance of unforeseen events.
Problem
The person in charge of managing the team makes an estimate and planning of the work to develop a project, including certain time buffers to be able to withstand possible unforeseen events. However, the appearance, or not, of a large number of circumstances not originally foreseen make it impossible to fulfil the initial planning in an optimised way.
Benefit
The development of an intelligent system capable of regenerating planning in the face of unforeseen events allows optimal use of resources to optimize delivery times.
METHODOLOGY & results
Architecture: On-premise development using local databases such as MySQL
Developing language: Java and Python
Data mining frameworks: KDD
ML techniques: Supervised learning algorithms (ANN, Decision Trees), unsupervised learning algorithms (HCA) and NLP techniques.
Results: Depending on the number and seriousness of the unforeseen events that arose during the development of different projects, the system allows compliance of between 77% and 90% of the deadlines established in the initial plan.
