arrow_back Optimising Model performance using automated ML pipeline for predicting purchase propensity @ Fractal Analytics
Application of machine learning in oil and gas industry
Submitted by Priyanka Raghavan (@priyankaraghavan) on Tuesday, 25 April 2017
Crisp talk for data engineering track
This talk describes the various machine learning algorithms used in the public SEG (Society of Exploration Geophysicist) challenge held in December 2016 to identify lithofacies based on well log measurements. Lithofacies are the different rock layers encountered during drilling, which are used to characterize the sub-surface. Correct classification of lithological facies helps in identifying target areas for hydrocarbon (oil and gas) extraction. The challenge concentrated on open-source tools, such as python and jupyter notebook, with collaborative code sharing via GitHub. The test data was provided by the SEG and the final submitted entry was one vs one multi class algorithm using Gradient boosting. The talk will explore the dataset, explain dataset conditioning, results on various algorithms used to train classifier, lessons learnt and future work.
The agenda will be as follows:-
1. Describing problem
2. Exploring dataset and conditioning
3. Algorithms used for best result
4. What did the winning team do
5. Lessons learnt and future work
Priyanka Raghavan is Team Lead on Prestack seismic Interpretation product at Schlumberger. She has worked with Schlumberger for the past 11 years in different products related to software used in seismic exploration. She started her career in Houston, followed by a stint in Germany before coming to Bangalore. She holds a Master degree in Software Engineering from Carnegie Mellon University. Her areas of interest are software architecture, design patterns and code optimization. In her free time you’ll see her honing her carnatic music skills.