Distributed collaborative prediction: Results of the machine learning contest

Abstract

<jats:p> The Geophysical Tutorial in the October issue of The Leading Edge was the first we've done on the topic of machine learning. Brendon Hall's article ( Hall, 2016 ) showed readers how to take a small data set — wireline logs and geologic facies data from nine wells in the Hugoton natural gas and helium field of southwest Kansas ( Dubois et al., 2007 ) — and predict the facies in two wells for which the facies data were not available. The article demonstrated with 25 lines of code how to explore the data set, then create, train and test a machine learning model for facies classification, and finally visualize the results. The workflow took a deliberately naive approach using a support vector machine model. It achieved a sort of baseline accuracy rate — a first-order prediction, if you will — of 0.42. That might sound low, but it's not untypical for a naive approach to this kind of problem. For comparison, random draws from the facies distribution score 0.16, which is therefore the true baseline. </jats:p>

A machine learning approach to facies classification using well logs

“…Our algorithm achieves an f-score of 0.61 averaged over the available 10 folds. This is perfectly in line with the f-score of about 0.62 we obtained on completely unknown wells according to Hall and Hall (2017), thus confirming the good generalization capability of the proposed strategy. Moreover, we computed also the f-score achieved without using feature augmentation.…”

“…Finally, as suggested by Hall (2016), we also evaluated our algorithm in terms of f-score. Hall and Hall (2017) declare that random guess score is around 0.16 due to class unbalancing.…”

“…This means, generate new features starting from the available ones, in order to better separate classes into an augmented feature space. This step was recognized also in Hall and Hall (2017) to be of great impact for the facies classification challenge.…”

“…On the same occasion, he announced a machine learning contest inviting readers to propose alternative solution for the same task. Results of the machine learning contest are resumed in a paper written by the organizers (Hall and Hall, 2017) and all submitted code is available online * .…”

“…In this paper we describe the machine learning pipeline we used for our submission to the facies classification challenge, highlighting the main characteristics that enable our method to achieve promising results. Specifically, our main contribution, as recognized by the organizers (Hall and Hall, 2017), has been a feature augmentation framework which has been subsequently used by all top scored teams.…”