Data Science JupyterLab environment and the Accelerated Data Science SDK are enhanced
- Services: Data Science
- Release Date: March 19, 2020
The JupyterLab notebook session interface is enhanced so that the JupyterLab environment now supports:
- The Variable Inspector extension.
- The Table of Content extension.
- Renders
geojson
,fasta
,plotly
,bokeh
, andjson
files natively. - Latex is rendered in markdown cells.
The Accelerated Data Science (ADS) SDK is enhanced as follows:
- Performance improvements of the ADW dataset loader resulting in a significant speedup.
DatasetFactory.open() with format='sql'
no longer requiresindex_col
to be specified. Additionally, thetable
parameter can be either atable
or asql
expression inDatasetFactory.open().
- ADS no longer instantiates an H2O cluster on behalf of the user. Instead the user needs to use
import h2o
to own and start their own cluster. - You can now profile Dask operations in ADS, which allows you to monitor the CPU and memory consumption in your notebook session.
- Dask version upgrade (version 2.10.1) with support for OCI Object Storage.
- Several bugs in ADS have been fixed including the upsampling recommendation, visualizations issues for model evaluation, and so on.