Demo notebooks

These notebooks give examples of how to run different parts of predictatops’s sequence of steps.

To a large extent the notebooks either run modules with the “_runner” ending in their name that directly execute code or they mirror the code execution patterns in the “_runner” modules.

1. The notebook Example_Every_Step_Via_HighLevel_Runner_Scripts_v1.ipynb. covers:

Running the high-level runner modules for each step one by one. This does the whole pipeline from data load, data evaluation, data processing, feature creation, model training, to prediction.

• fetch_demo_data.py

• configurationplusfiles_runner.py

• checkdata_runner.py

• load_runner.py

• split_runner.py

• wellsKNN_runner.py

• features_runner.py

• balance_runner.py

• trainclasses_runner.py

• predictionclasses_runner.py

---

NOTE: The following notebooks don’t use “_runner” modules. Code is imported from predictatops and more code written in the notebooks themselves expresses how and when to execute that code.

2. The notebook Example_firstSteps_modules_fetchdata_configuration_checkdata.ipynb. Covers the initial data load, data evaluation, and configuration. Specifically it covers the modules:

• fetch_demo_data.py

  • Gets the data into the data folder from the demo folder in the repository.

• configurationplusfiles.py

  • Instantiates the class objects that contain information on how to run the rest of the program as it applies to input files, output files, and general configuration.

• checkdata.py

  • Helps to find out which wells have the tops and curves you need or the inverse to find out which curves you have if you want a certain number of wells in your data population.

3. This notebook TBD

• load.py

  • Loads the data from wells identified in the checkdata.py step above.

• split.py

  • Splits the data from the load.py step into train and test groups.

4. The notebook `Example_module_wellsKNN_v1.ipynb https://github.com/JustinGOSSES/predictatops/blob/master/demo/Example_module_wellsKNN_v1.ipynb/>`_. covers:

• wellsKNN.py

  • Uses well location to identify wells next to one another up to X number of neighbors.

5. The notebook Example_module_balance_v1.ipynb. covers:

• balance.py

  • balancing the populations of classes, or labels, associated with each depth point in each well for training, so there are roughly equal number of classes.

6. The notebook Example_module_features_and_balance.ipynb. covers:

• features.py

  • creates features not already created in wellsKNN.py

• balance.py

  • balancing the populations of classes, or labels, associated with each depth point in each well for training, so there are roughly equal number of classes.

7. The notebook Example_module_trainclasses_v2.ipynb. covers:

• trainclasses.py

  • training the model

8. The notebook Example_module_predictionclasses_v1.ipynb. covers:

• predictionclasses.py

  • making the top prediction