Rapid region wide production forecasting

1. A method comprising:

identifying base data of a well in a plurality of wells of a region;

selecting, using the base data and from a set of a models comprising a rich machine learning model, a location based machine learning model, and a decline curve model, a well model;

generating, based on the selecting, a forecasted production of the well using the base data and the well model;

aggregating a plurality of forecasted productions of the plurality of wells, the plurality of forecasted productions including the forecasted production, to generate a region forecast by applying each of the rich machine learning model, the location based machine learning model, and the decline curve model to the base data; and

using the region forecast to perform a field operation.

2. The method of claim 1 , further comprising:

generating the forecasted production of the well by:

when the base data includes production data for the rich machine learning model and an end date of the base data is before a threshold number of months of production, using the rich machine learning model to generate the forecasted production to the threshold number of months of production, and then using the decline curve model to generate the forecasted production from the threshold number of months of production to a forecast date.

3. The method of claim 1 , further comprising:

generating the forecasted production of the well by:

when the base data does not include production data and end date of the base data is before a threshold number of months of production, using the location based machine learning model to generate the forecasted production to the threshold number of months of production, and then using the decline curve model to generate the forecasted production from the threshold number of months of production to a forecast date.

4. The method of claim 1 , further comprising:

generating the forecasted production of the well by:

when the base data has an end date after a threshold number of months of production, using the decline curve model to generate the forecasted production to a forecast date.

5. The method of claim 1 , wherein the rich machine learning model uses one of a gradient boosted random forest algorithm and a random forest algorithm.

6. The method of claim 1 , wherein the location based machine learning model uses one of a gradient boosted random forest algorithm and a random forest algorithm.

7. The method of claim 1 , further comprising:

ingesting training data of a plurality of training wells of a target region;

training the rich machine learning model on the training data, the training data comprising rich data from wells in the target region; and

training the location based machine learning model with the training data.

8. The method of claim 7 ,

wherein ingesting the training data further comprises:

removing training data for wells that have less than a threshold number of months of production; and

removing training data for wells that include anomalous data patterns, including patterns that deviate from an exponential decline by a threshold amount.

9. A server system comprising:

a processor;

a memory coupled to the processor; and

the memory comprising an application that executes on the processor, uses the memory, and is configured for:

identifying base data of a well in a plurality of wells of a region;

selecting, using the base data and from a set of a models comprising a rich machine learning model, a location based machine learning model, and a decline curve model, a well model;

generating, based on the selecting, a forecasted production of the well using the base data and the well model;

aggregating a plurality of forecasted productions of the plurality of wells, the plurality of forecasted productions including the forecasted production, to generate a region forecast by applying each of the rich machine learning model, the location based machine learning model, and the decline curve model to the base data; and

using the region forecast to perform a field operation.

10. The server system of claim 9 , wherein the application is further configured for:

generating the forecasted production of the well by:

when the base data includes production data for the rich machine learning model and an end date of the base data is before a threshold number of months of production, using the rich machine learning model to generate the forecasted production to the threshold number of months of production, and then using the decline curve model to generate the forecasted production from the threshold number of months of production to a forecast date.

11. The server system of claim 9 , wherein the application is further configured for:

generating the forecasted production of the well by:

when the base data does not include production data and an end date of the base data is before a threshold number of months of production, using the location based machine learning model to generate the forecasted production to the threshold number of months of production, and then using the decline curve model to generate the forecasted production from the threshold number of months of production to a forecast date.

12. The server system of claim 9 , wherein the application is further configured for:

generating the forecasted production of the well by:

when the base data has an end date after a threshold number of months of production, using the decline curve model to generate the forecasted production to a forecast date.

13. The server system of claim 9 , wherein the rich machine learning model uses one of a gradient boosted random forest algorithm and a random forest algorithm.

14. The server system of claim 9 , wherein the location based machine learning model uses one of a gradient boosted random forest algorithm and a random forest algorithm.

15. The server system of claim 9 , wherein the application is further configured for:

ingesting training data of a plurality of training wells of a target region;

training the rich machine learning model on the training data, the training data comprising rich data from wells in the target region; and

training the location based machine learning model with the training data.

16. The server system of claim 15 , wherein ingesting the training data further comprises:

removing training data for wells that have less than a threshold number of months of production; and

removing training data for wells that include anomalous data patterns, including patterns that deviate from an exponential decline by a threshold amount.

17. One or more non-transitory computer readable mediums comprising computer readable program code for:

identifying base data of a well in a plurality of wells of a region;

selecting, using the base data and from a set of a models comprising a rich machine learning model, a location based machine learning model, and a decline curve model, a well model;

generating, based on the selecting, a forecasted production of the well using the base data and the well model;

aggregating a plurality of forecasted productions of the plurality of wells, the plurality of forecasted productions including the forecasted production, to generate a region forecast by applying each of the rich machine learning model, the location based machine learning model, and the decline curve model to the base data; and

using the region forecast to perform a field operation.

18. The one or more non-transitory computer readable mediums of claim 17 , further comprising computer readable program code for:

generating the forecasted production of the well by:

when the base data includes production data for the rich machine learning model and an end date of the base data is before a threshold number of months of production, using the rich machine learning model to generate the forecasted production to the threshold number of months of production, and then using the decline curve model to generate the forecasted production from the threshold number of months of production to a forecast date.

19. The one or more non-transitory computer readable mediums of claim 17 , further comprising computer readable program code for:

generating the forecasted production of the well by:

when the base data does not include production data and an end date of the base data is before a threshold number of months of production, using the location based machine learning model to generate the forecasted production to the threshold number of months of production, and then using the decline curve model to generate the forecasted production from the threshold number of months of production to a forecast date.

20. The one or more non-transitory computer readable mediums of claim 17 , further comprising computer readable program code for:

generating the forecasted production of the well by:

when the base data has an end date after a threshold number of months of production, using the decline curve model to generate the forecasted production to a forecast date.