The simplest way to query your data is with
You can also upload data to the cloud. Non-production accounts may have limitations to the size of the uploaded data.
Soil accepts typical python structures like dicts and lists as long as they are json serializable and numpy arrays and pandas data frames.
You can pass metadata as a dictionary (json serializable) to the uploaded data.
By default soil recognises basic data types that can be passed in soil.data. These are:
To access the data you can do:
The schema of the metadata can be different for every data structure type.
You can pass parameters if the data structure supports it:
soil.alias(name, ref, roles=None) allows to easily reference the data. This can be useful in a continuous learining environment for example.
- roles is an optional list of strings that indicate which users will have permissions to read the result. At least one role must coincide with one of the roles of the user.