In a governed lakehouse, what is a practical advantage of doing both exploratory and predictive work against the platform's own tables rather than on extracted copies?
Select an answer to reveal the explanation.
Short Explanation
Working in place means your exploration and model read the same governed, versioned tables. You can trace the lineage and reproduce the result later, and you don't end up with an extracted copy that drifts or slips outside policy.
Full Explanation
The correct choice is the option stating that both stages read the same governed, versioned data, so findings and models rest on a lineage that can be traced and reproduced. Keeping analysis on governed tables preserves lineage: you can identify what data a finding or a model was built from, reproduce it, and know that access policy applied throughout. Extracted copies break all three - they drift from the source and sit outside the policies that governed the original.
The option claiming this is the only way to run statistical functions, because they are unavailable outside the platform, is incorrect because statistical functions are widely available in other tools. The advantage is governance and reproducibility, not exclusive capability.
The option claiming this guarantees models will be more accurate, since governed data is inherently cleaner, is incorrect because governance concerns access, lineage, and policy rather than statistical quality. Governed data can still be incomplete or skewed, which is why exploration remains necessary.
The option claiming this removes the need to validate models, because the platform validates the data, is incorrect because platform governance says nothing about whether a model generalises. Validation remains the modeller's responsibility.
For the exam, associate in-place analysis with lineage, reproducibility, and continuous policy enforcement.