2023-10-29
Check out the complete article HERE.
Feature engineering is the process of transforming raw data into meaningful and usable features, which is as critical as making sure a goat is well fed to ensure it thrives. Machine learning pipelines can be built primarily using TensorFlow or PyTorch, two well-known frameworks in the field. In this discussion, I will summarize the types of Feature Engineerinng and present differences between these two platforms in terms of feature engineering capabilities and approaches.
For better illustration, I will use the California housing dataset, which contains U.S. Census 1990-derived housing data. The nuances of each framework can be easily clarified through specific code examples.
The TensorFlow code I reference is adapted from Google Cloud. On the other hand, I wrote my own PyTorch code that is structurally and functionally consistent with the TensorFlow examples to ensure a comparable implementation.