Continuing Education Course With Guest Speaker Whitney Trainor-Guitton

Date/Time
Date(s) - 08/07/2019 - 08/08/2019
8:00 am - 4:45 pm

Location
410 17th Street, 2nd Floor

Categories No Categories

Please Join Us August 7th and 8th

For a Continuing Education Course on Data Analytics with Python Scikit Learn

With Guest Speaker Whitney Trainor-Guitton

Located at 410 17th Street 2nd floor Conference Room Denver

8-4:45pm

Includes breakfast and lunch both days.

Who Should Attend

Geoscientists, engineers and GIS professionals charged with interpreting data and who are interested in statistical techniques to find important relationships in their play. Practioners that are interested in “looking under the hood” at some of the algorithms in proprietary software.

Course Description

In this 2-day short course, participants will become familiar with statistical learning methods specifically useful for petroleum industry data sets. We will work with real world data such as log and production data from Marcellus shale and the Jonah gas field. Course may be modified to provide Python fundamentals if interest is expressed.

Learning outcomes:

• Import and visualize data in Python Jupyter notebook
  • Histograms, scatterplots, time-series.
• Apply toolkits to preprocess datasets for input to statistical and machine learning algorithms, including methods of feature extraction and dimensionality reduction (principal component analysis).
• Apply statistical and machine learning toolkits to geoscience datasets, including applications of Time series prediction, classification, and clustering.
• Explain in words the necessity for model validation and apply model evaluation in Python via the sci-kit learn library.
  • Calculate and describe in words the difference between the different metrics
  • Describe why splitting of data into training and testing is necessary
  • Describe and implement diagnostics for testing for under-fitting or over-fitting

Required Hardware

Each participant is expected to bring their own laptop for which they have privileges to install the relevant software: Python 3.7. Instructions and guidance will be provided for installing and accessing Python Jupyter notebooks, and specifically installing and utilizing the scikit-learn API.

Preliminary Schedule

Day 1

• 08:00 Welcome & introductions

• 08:30 Demonstration: Introduction to Jupyter notebooks
• 09:00 Data Import Exercise: Read in decline curve, logs, and production data
• 09:45 Presentation on Important Libraries: Numpy & Matplotlib
• 10:00 Break
• 10:15 Python practice: arrays and visualizations
• 12:00 Lunch
• 13:00 Morning recap: Got’s and Need’s
• 13:30 Python practice: Exploratory Data Analysis with preferred geologic data
• 14:00 Demonstration & Practice: Data selection (Pandas)
• 15:00 Break
• 15:15 Activity: Random Forest for prediction
• 16:00 Activity recap: Machine Learning principals
• 16:45 Wrap up: Got’s and Need’s from Day 1

Day 2

• 08:00 Activity: Categorizing prediction problems (supervised/unsupervised, classification/regression)
• 08:30 Presentation on Important Libraries: Sci-Kit Learn
• 09:00 Python practice: Applying a model from Sci-Kit Learn (Random Forest: production prediction from well logs/attributes)
• 10:00 Break
• 10:15 Presentation: Unsupervised methods for exploratory data analysis
• 11:00 Python practice: Scaling, Normalization, Principal Component Analysis
• 12:00 Lunch
• 13:00 Morning recap: Got’s and Need’s
• 13:30 Presentation: ARIMA model(s) for time series data
• 14:30 Python practice: Decline curve analysis
• 15:00 Break
• 15:15 Presentation: Naïve Bayes
• 15:45 Python practice: Production Classification with Naïve Bayes
• 16:45 Wrap up

Course Instructor

Whitney Trainor-Guitton is an assistant professor of Geophysics at Colorado School of Mines, where she works on stochastic model building techniques utilizing geophysical data. Whitney uses deep learning for evaluating the information content in seismic migration images. Whitney completed her master’s degree in geophysics and her PhD in the interdisciplinary Program of Earth, Energy and Environmental Science, both from Stanford University. For her dissertation, she developed transferable value of information (VOI) methodologies for spatial earth problems.

Bookings