search

chews0n / glowing-waffle

SPE Calgary Data Science Mentorship program 2021
Apache License 2.0
1 stars 2 forks source link

Visualize the Resutls #17

Open chews0n opened 3 years ago

chews0n commented 3 years ago

Create a way to plot your features and essentially give an end report of your findings from the model. It would be best to have multiple graphs and all of the information (including the input) in one file in order to make it easy for either replicating runs or varying your runs with something like retraining of the model.

You guys had lots of great visualisations and outputs from the initial assignment (saved in the docs --> practice-problem --> ) folders, try to recycle and repurposed some of that to make your live's easier.

chews0n commented 3 years ago

List of Tables, Values, and Plots to return to the User:

  1. Feature Importance - Tornado Plot
  2. Actual Vs. Predicted - Well Number vs. Predicted Production - This should live in the training of the model
  3. Partial Dependency Plots
  4. IP90/IP180 - we can return a value for the result as they're not really time dependant
  5. Have some sort of confidence interval, so maybe you do have multiple models that are trained with different sets of data, let's leave that until the end if we get there.
  6. Separate visualizations for feature importance
  7. Gas vs. Oil Wells
  8. Heat Map of multiple case runs