Event Series Prediction as decision support system at fast paced processes
Paper in proceeding, 2016
Prediction is used in sciences, technologies, and daily life. This paper investigates how prediction influence three fast paced creative processes. The setup for the experiments are described and the participants report on their experiences. Based on the experiences from these three types of processes, the article discuss how a design or alteration process may be guided by predictions. For prediction we present the algorithm 'Event Series Prediction' (ESP), and account for the ways it can be implemented in any real time, fast paced process. ESP algorithm provides, based on the current and past scope of events, representations pieced together from its memory. These representations achieve similarities with the future events in the environment. However, changing the environment based on such prediction will in turn change the outcome of events in the environment. These consequences are sought to be demonstrated through experimentation. Once understanding the mechanism, our goal is to scale the temporal dimension of experiments to challenging, time-consuming tasks such as modifying the built environment. Scaling the process in temporality is shown possible, and proves to heighten the relevance and performance of the system, in that the human memory is outperformed by the system when using longer temporal scale.
Machine learning
Prediction
Artificial intelligence
Forecasting
Architecture
Modification
Environment