Clean tabular dataset
You can’t train an experiment if the dataset has rows with empty values or malformed rows. The platform includes a tool that will clean your tabular dataset and make it usable.
Remove rows with empty values
The input dataset can’t contain any blank values. This makes sense - the model doesn’t know how to parse a blank value. The cleaning tool will remove rows with empty values in your input data.
Don’t clean too much - keep as much data as you can
Deselect the features that you won’t need in your experiment. This way, you keep as much data as possible.
When you’re working with deep learning, you want to train on as much data as possible. That’s why it’s a good idea to keep as much data as possible.
Store name, Advertising level, or Holiday is of no interest in your experiment. If you deselect these features, then it doesn’t matter if a row has an empty value for these features. The row will be kept for use in the experiment.
Correct csv format
CSV files are known to have different formats depending on the tool you are working with. The Peltarion platform requires a certain format, so the tool will make sure the file complies with our requirements.