TIDE – Toyota Intelligent Data Exploration

One of the main limiting factors will be the duration of the project, in particular the data collection part. For example, for tracking of how does the vehicle behaviour and condition change over time – it is not clear how much of that will we be able to actually observe during the half a year pilot. Therefore, we will focus on methods for discovering such trends and propose approaches that are suitable for forklift environment, but it is likely that no significant changes will occur in such a short time. Since such wear patterns are very interesting to investigate, we will consider performing several controlled experiments – probably quite short and in limited set of conditions – on different types of forklifts...

Nowadays, data is easier to create, store and handle than ever. It can be used to great benefit, however, we note that it is not the data itself that is interesting, but rather the underlying reality. The data is only useful in as far as it can help in understanding and drawing conclusions about that reality, which is the foundation of “Prescriptive Analytics”. The wealth of knowledge hidden in Big Data is often not exploited to the full potential because of how difficult it is to make the connection between the data and the reality, especially for users who are not Data Scientists.

We have identified five areas of interest in the project: wear patterns; repair effects & technician support; self-monitoring, survival analysis and breakdown prediction; prediction management; and abnormal data. They are all related to using available data for increased understanding of the product and its usage, however, they approach this goal from different points of view. In this exploratory phase it is important to try different methods, based both on how interesting and how feasible they are, given the time available.