Using machine learning techniques to predict compressional travel-time (DTC) in an offset well
Purpose: Predicting acoustic slowness data from triple compbo logs based on ML model that was built on offset well logging data. SB16 well is used to build ML model. The model is then deployed to predict slowness data in SB17 offsett well. This section includes data exploring, cleaning, correcting, completing, creating, converting, and deploying ML technique.
Data: The data used is available at https://certmapper.cr.usgs.gov/data/PubArchives/OF00-200/WELLS/WELLIDX.HTM Provided data is part of the USGS Open File Report 00-200 public dataset.
Tools: Python, Jupyter Notebook, Pandas, Numpy, Lasio, Seaborn, Xgboost
Improvements: Outlier Detection, ML Hyper parameter optimization
Credits: Some scprits from work of Matteo Niccoli, Mihai, Matt Hall was used.