Bridging the Gap between Data and Value

Transcription

1 Bridging the Gap between Data and Value Machine Learning in Oil & Gas Energy Conference Network Bill Robertson Senior Manager, Enterprise Analytics April 19, B A K E R H U G H E S I N C O R P O R A TED. A LL R I G H TS R E S E R V E D. TERMS A N D C O N D I TI O N S O F U S E : B Y A C C E P TI N G THIS DOCUMENT, THE RECIPIENT A G R E E S THAT THE DOCUMENT TOGETHER W I TH A LL I N FORMATI O N I N C LUDED THEREIN I S THE C O N FI D E N TI A L A N D P R O P R I E TARY PROPERTY OF B A K E R H U G H E S I N C O R P O R A TED AND INCLUDES VALUABLE TRADE SECRETS AND/OR PROPRIETARY INF ORMA TI O N O F B A K E R H U G H E S (C O LLECTI V E LY "I N FORMATI O N "). B A K E R H U G H E S R E TAINS A LL R I G H TS UNDER COPYRIGHT LAW S A N D TRADE SECRET LAW S O F THE UNITED S TATES OF A M E R I C A A N D O THER COUNTRIES. THE RECIPIENT FURTHE R A G R E E S THAT THE DOCUMENT M A Y N O T B E D I S TRIBUTED, TRANSMITTED, C O P I E D O R R E P R O D U C E D I N W H O LE OR IN P A R T B Y A N Y M E A N S, E LECTRONIC, MECHANICAL, O R O THERWISE, W I THOUT THE E X P R E S S P R I O R W R I TTEN C O N S E N T O F B A K E R H U G H E S, A N D M A Y N O T B E U S E D D I R E C TLY OR INDIRECTLY I N A N Y W A Y D E TRIMENTAL TO BAKER HUGHES INTEREST.

2 Enterprise Analytics Multi-disciplinary team of data analysts, domain experts, and application developers Over 120 years of industry experience Based in Houston, Texas Successful record delivering solutions from exploration to production optimization 2

3 Data is everywhere 90% of the data in the world today has been created in the last two years 3

4 Data is an asset Analytics will define the difference between the losers and winners The average initial increase in profits from big data investments was 6% that increased to 9% for investments spanning five years. 4

5 Leveraging data is hard Challenges Which data should you use? How should you handle analytics? How can you use the insights you ve gained to transform your operations? 5

6 Data analysis breakdown Data preparation accounts for almost 80% of the work of data scientists 6

7 Today s enterprise Functional Groups HR Finance HSE Quality Legal Supply Chain Product Lines Operations Enterprise Analytics Data Governance. 7

8 Data analysis evolution 8

9 Analytic goals Goal Improve understanding of customers 33% 45% 46% Retaining customers 30% 36% 37% Improving customer experiences 22% 36% 36% Selling products / services to existing customers 23% 33% 35% Market research / survey analysis 29% 36% 34% Acquiring customers 23% 32% 32% Improving direct marketing programs 22% 27% 30% Sales forecasting 19% 27% 27% Fraud detection or prevention 21% 23% 26% 9

10 Challenge the business case Stakeholder Analyst Do you know who our top customers are in region X? Why? I need to know what our revenue distribution looks like. Why? Last quarter s revenues are lower than expected. Why? I don t know, but I think we can increase our share of wallet by 5%.??? 10

11 Predictive maintenance for Artificial Lift When will the electronic submersible pump (ESP) fail? If we can predict this, where s the potential value? Considerations Land Offshore Lost production $$$ $$$ Rig availability $ $$$ Materials inventory $ $$$ Alternate suppliers $ $$$ Potential value $$ $$$$$ 11

12 Barriers to success What is considered a failure? How far in advance can we make an accurate prediction? Do we have historical data for the intended use case? What s the quality of the data? Is there management support for implementing a change? 12

13 Change management Virginia Satir Change Process Model 13

14 Product selection for Flow Assurance Given a fluid sample, which products will mitigate the production risks? Fluid Chemistry System Conditions f Which products will perform well? Before - Compatibility tests - Static bottle tests - Kinetic turbidity tests After (i.e., use a prediction) - Dynamic tube-blocking tests 14

15 All data are not created equal Structured Unstructured Low Cost of Ownership High 15

16 Evaluating classifier performance 95% Lab Test 73% Clear True Positive False Positive 61% Probability of Passing 49% 37% Threshold? Not Clear 25% False Negative True Negative 13% 1% Clear Actual Not Clear 16

17 Data management Manage your data well so that good analyses are possible Pre-requisites Value Acquire data Process and analyze data Move data Store data 17

18 Takeaways Ensure your business cases are well-defined Manage your data Empower yourself Bill Robertson Senior Manager, Enterprise Analytics Baker Hughes