Hi,
Im not sure of how will I proceed with data-centric modelling.
I assume this is the sequence of steps to follow in data-centric modelling
- Create a base model(assume model using logistic reg) with the existing data. The base model is not fine tuned.(should I fine tune the model ??)
- on the base model, try and improve the data using data augmentation ,error analysis .
- after step2, assume base model gives satisfactory f1 score.
But what if I want to try various other algorithm other than logistics reg, how should I proceed, should I repeat step1-3 on the fine-tuned dataset or should I run the new model on original dataset.
Kindly advise