Explore the potential of data innovation for your company with the EluciDATA Starter Kits

In collaboration with the EluciDATA user group, a number of concrete industrial use cases have been identified in the domains 'entity profiling and recommendation' and 'predictive analytics and forecasting'. For each of these use cases, a starter kit is being developed, to illustrate the potential of data innovation for a specific use case, in this way allowing companies to start with data innovation at a faster pace. As an example, a starter kit focusing on 'resource demand forecasting' is now available. 

While ever more data is available, the technology to exploit that data is in the process of maturing every day. The opportunities created by data are manifold and widely recognised, its potential is still underutilised in most companies. Therefore, two years ago, Sirris initiated the EluciDATA project, focusing on accelerating data innovation. The goal of this project is to support companies in exploiting their data by analysing collective needs and challenges and investigating solutions using existing technology. The project’s relevance is witnessed by the substantial interest in the user group, consisting of more than 30 companies from a wide variety of domains (e.g., manufacturing, energy, mobility, marketing), including both problem owners and technology providers. 

Two core domains

In the first phase of the project, we elicited needs and challenges that companies in the user group are facing. Based on this input, two core domains in which data innovation could be instrumental were identified:

  • entity profiling and recommendation is concerned with grouping entities (i.e., users, machines, assets, etc.) together according to similar characteristics, and extracting useful recommendations from these profiles.
  • predictive analytics and forecasting, in which historical data from monitoring in the past is analysed in order to predict the future. This also encompasses identifying recurring patterns in the data as well as unexpected behaviour (e.g., a failure of a machine, a decrease in physical activity when monitoring health parameters, etc.). 

Starter kits

In collaboration with the user group, a number of concrete industrial use cases have been identified in each of these two domains. For each of these use cases, a starter kit is being developed in the course of the next two years. The goal of these starter kits is to illustrate the potential of data innovation for a specific use case, and in this way allow companies to start with data innovation at a faster pace

As an example, a starter kit focusing on resource demand forecasting is available. This is relevant for a number of companies in a variety of domains:

  • Forecast parking demand or traffic density in a particular neighbourhood, in order to control traffic in intelligent ways (e.g. divert to alternative routes or parking spots).
  • Predict battery consumption based on operating conditions, in order to intelligently suggest recharging moments or switch to low-power mode
  • Estimate cost of consumables (electricity, gas, ink, paper, ...) based on usage data, in order to deliver exactly on time or minimise total cost.

As resources in many domains, systems and applications are scarce, knowing in advance what the demand will be can help to plan ahead and guarantee that a sufficient amount is available when needed and to avoid costly countermeasures in case of shortage. 

The overall goal of this starter kit is to demonstrate how to forecast the demand for a particular resource at a particular point in the future. It illustrates the complete process of transforming the raw data into an accurate forecast, using a number of publicly-available software tools and libraries. This involves several steps:

  1. Identifying the relevant data sources : in most cases, companies gather data from different sources simultaneously. In the first step of this starter kit, we explain how to identify which sources are relevant for resource demand forecasting and how to combine data from multiple heterogeneous sources.
  2. Data exploration and feature extraction : A second step in this process is to explore the characteristics of the data, in order to find interesting patterns, trends or anomalies. In the starter kit, several ways on how to do this are explained by means of illustrative examples. Based on the resulting observations, so-called features are defined, i.e. characteristics of the data that are helpful in forecasting the resource demand (e.g., the consumption in the previous day or week, the habits of the consumer, etc.)
  3. Data analysis and evaluation : the defined features serve as input for an algorithm implementing a forecasting model. The starter kit shows several possible models and explains how to evaluate the quality of the resulting predictions. 

Next to the starter kit on resource demand forecasting, several other starter kits are available or under development, dealing with among others:

  • The analysis of the interactions a user has had with a particular product, in order to extract recommendations and improve the user support.
  • Quality improvement through dynamic optimisation of machine settings.
  • Techniques for (advanced) data visualisation. Feature engineering, explaining the process of transforming the data and extracting the most relevant distinguishing characteristics out of it. 

Complementary to these Starter Kits, we also offer a number of interactive training sessions on these topics. You can find further information and the possibility to register in our agenda

Are you interested in exploring the potential of data innovation in your own company’s context, using the Starter Kits? Contact us for further information.