Linde.Intel.AI.Challenge 2018 is a conference dedicated to solving the future of smart logistics with machine learning and AI in a hackathon - a competition to develop, in teams in a limited amount of time, innovative solutions.
Topics
- Assessment and development of a multi-time series approach (is possible to learn across different customers instead of customer-individual model training? Is it possible to identify clusters/segments of storage vessels whose consumption patterns and behaviour is similar?)
- Creation of prediction algorithms (RNN based approaches preferred!) for liquid gas tank levels (one day ahead and multi-day forecasts on an hourly basis) of our bulk customers
- Interactive dashboard
- Inclusion of relevant external data (customer individual information (industry, location, weather etc.)
- Clusters of customers/industries/common consumption patterns across groups
- Explain the influence of individual features
