Session 6: An Analytical Approach to Data Preparation for Machine Learning: A 5000 Well UK North Sea Case Study
Date & Time
Wednesday, May 20, 2020, 2:00 PM - 2:30 PM
The real-world application of Machine and Deep Learning techniques within the E&P industry is hampered by the nature of the legacy data. The UKCS is a mature basin and the existing data reflect the range of technologies, techniques and data formats. Historically weaker governmental legislation on data storage and retention has led to unprecedented data challenges, particularly relevant in the preparation and conditioning of data for applying Data Science methodologies which are very much on-trend across the sector. In partnership with the UK OGTC, DataCo has collaborated with the top 10 acreage holders in the UK Northern North Sea, the UK OGA, the Norwegian NPD and two Data Science partners to act as the catalyst in the preparation of legacy data for the identification of zones of overlooked missed pay. DataCo’s preparation of 500 wells consisting of more than 9000 wellbores, using machine learning, big data technologies and cloud computing, will make available a machine-readable, data science ready, data-set consisting of Automatically Spliced JWLs from more than 2,000,000 individual files and conditioned Deviation Surveys, Well Headers, Well Formation Tops, CCC and RCA data.