When input and target are selected, not all algorithms will be able to consume this data. Same goes for models. If the information about the field datatypes is kept in the dataset info table, and acceptable input types for algorithms and models are also provided, a user can be presented only by the compatible choices, so there is much less of a chance to mess things up.
Needs to be done:
[ ] Make fields for Datasets, Models and Algorithms in the boonai/boonai/model.py to store info about the actual field type (for Datasets) and allowed field types (for Models and Algorithms).
[ ] Make a query get only compatible algorithms when fields are selected (for training).
[ ] Make a query get only compatible models when field is selected (for predicting).
When input and target are selected, not all algorithms will be able to consume this data. Same goes for models. If the information about the field datatypes is kept in the dataset info table, and acceptable input types for algorithms and models are also provided, a user can be presented only by the compatible choices, so there is much less of a chance to mess things up.
Needs to be done:
boonai/boonai/model.pyto store info about the actual field type (for Datasets) and allowed field types (for Models and Algorithms).