pandas dataframe versioning 100x times faster
In omega-ml pandas dataframes can be stored as versioned change-sets. This means you can get back a previous-version at any point in time or by a tag. Changes are tracked by index so you get to choose the key. Previously this was very slow, it is now 100x times faster. How? Smartly collecting all database updates and doing them in one go, instead of one-by-one.