Closed blackadar closed 4 years ago
The data processing steps need to be packaged cleanly into a pipeline in order for the project to accept new data as input. Any cleaning and segmentation done with the training data needs to be mirrored on the implementation.
Pipeline will at first be separate from the model, and the output can be saved to save time on preprocessing.
The data processing steps need to be packaged cleanly into a pipeline in order for the project to accept new data as input. Any cleaning and segmentation done with the training data needs to be mirrored on the implementation.