if your data can fit into arrays, then there's no advantage in terms of the types of aggregations.
however mongo allows you to store complex structures, think nested dictionaries/lists, and query on those nested structures, even allowing you to reach inside of nested structures to do the querying.
I guess you could do nested structured arrays in numpy, I've never done that before.
I use h5py's datasets (which are organized hierarchically and stored in compressed chunks) to do basic filtering and then load a fraction of my data into memory as numpy arrays.
Comments
if your data can fit into arrays, then there's no advantage in terms of the types of aggregations.
however mongo allows you to store complex structures, think nested dictionaries/lists, and query on those nested structures, even allowing you to reach inside of nested structures to do the querying.
I guess you could do nested structured arrays in numpy, I've never done that before.
I use h5py's datasets (which are organized hierarchically and stored in compressed chunks) to do basic filtering and then load a fraction of my data into memory as numpy arrays.