Heterogeneous data inputs

[cmutel]

Imagine we have two dataframes in our data warehouse, and these dataframes have different shapes:

DataFrame A:

+----+------+------------+------------------+
|    | Well | Production | Fugitive Methane |
+----+------+------------+------------------+
|  0 | A142 | 42562      | 2543             |
|  1 | B427 | 29657      | 4852             |
|  2 | C492 | 85273      | 1673             |
+----+------+------------+------------------+

DataFrame B:

+----+------+------------+
|    | Well | Production |
+----+------+------------+
|  0 | D246 | 17349      |
|  1 | D247 | 93745      |
|  2 | D248 | 57263      |
+----+------+------------+

In DataFrame A, we have correlated observations. We want to keep these rows linked throughout the calculation. However, due to the differing shapes, we can't just stack the two DataFrames on top of each other (the shapes don't match).

We can add NaN or other dummy values to get the shapes to match, or assume that we have gap-filling models (though this isn't always desirable - sometimes we would like to know when we do and do not have real data).