Version 0.1.0 (Beta)¶
Welcome to the first release of VAST Orbit - Python data science for VAST DataBase!
Important
Beta Status: VAST Orbit 0.1.0 is in beta. Production-ready version 1.0.0 coming soon. API may change based on feedback. Report issues at `vastsupport.slack.com <https://vastsupport.slack.com>`__.
Release Highlights¶
Python Data Science for VAST
VAST Orbit 0.1.0 brings complete data science workflows to VAST DataBase - prepare, explore, analyze, and build ML models with in-database execution.
Core Features:
Data Preparation in VAST - Clean, transform, engineer features at any scale
Interactive Exploration - Charts and visualizations with intelligent sampling
400+ Functions - Complete analytics toolkit executing in VAST
10 ML Algorithms - Embedded models for training and inference
Multi-Source Access - Query VAST tables, files, and external databases
Zero Data Movement - All processing in VAST DataBase
What’s Included¶
Data Preparation:
fillna,dropna,drop_duplicates- cleaning in VASTnormalize,scale,encode- transformations in-databaseFeature engineering at petabyte scale
Statistical profiling and quality checks
Interactive Exploration:
Histograms, scatter plots, correlation matrices
Box plots, violin plots, KDE
Intelligent sampling for instant visualization
Statistical analysis (mean, median, variance, quantiles)
Analytics:
400+ functions executing in VAST
pandas-like DataFrame operations
Aggregations, joins, window functions
Time series and geospatial analysis
Machine Learning:
10 embedded models (RandomForest, GradientBoosting, LinearRegression, etc.)
sklearn model import support
In-database inference at scale
Production-ready deployment
Core Modules:
vastorbit.VastFrame- pandas-like DataFrame for VASTvastorbit.machine_learning.vast- ML algorithmsvastorbit.plot- Visualization libraryvastorbit.sql- SQL execution utilitiesvastorbit.stats- Statistical functions
Supported Platforms:
Python 3.12+
Linux and macOS
VAST DataBase 5.0.0-sp10 or later
Example Usage:
import vastorbit as vo
# Connect to VAST DataBase
vo.new_connection({
'host': 'vast-cluster.com',
'catalog': 'vast_catalog'
})
# Query data
vdf = vo.VastFrame('sales_data')
# Data preparation - all in VAST
vdf = vdf.fillna({'revenue': 0})
vdf = vdf.drop_duplicates()
# Explore with charts
vdf['revenue'].hist(nbins=20)
vdf.scatter(['sales', 'revenue'])
# Analyze
summary = vdf.groupby(['region'], ['sum(revenue) AS total'])
# Train ML model
from vastorbit.machine_learning.vast import RandomForestClassifier
model = RandomForestClassifier(n_estimators = 4)
model.fit(vdf, ['feature1', 'feature2'], 'target')
# In-database inference
predictions = model.predict(vdf)
Key Capabilities¶
In-Database Data Preparation:
Clean and transform data directly in VAST
Handle missing values, outliers, duplicates
Feature engineering at any scale
Statistical profiling and validation
Interactive Exploration:
Generate charts with intelligent sampling
Analyze distributions and correlations
Discover patterns and anomalies
Visualize billions of rows instantly
Multi-Source Analytics:
Query VAST tables and files
Access external databases (PostgreSQL, MySQL, MongoDB)
Join across sources
Unified Python API
In-Database ML:
10 embedded algorithms ready to use
Import sklearn models
In-database inference in VAST
Production-scale scoring
Beta Limitations¶
As a beta release:
API may change before 1.0.0
Documentation actively expanding
Some advanced features in development
Feedback welcome for improvements
Getting Started¶
Installation:
pip install vastorbit
Documentation:
Getting Started - Installation and setup
User Guide - Data preparation and analytics
Chart Gallery - Visualization examples
Machine Learning - ML workflows
API Reference - Complete API reference
Examples - Hands-on tutorials
Support:
Slack: vastsupport.slack.com
Roadmap to 1.0.0¶
Production Release Plans:
API stabilization based on beta feedback
Expanded documentation and tutorials
Additional ML algorithms
Enhanced data preparation functions
Advanced visualization capabilities
Performance optimizations
Production hardening
We’re excited to hear your feedback as we work toward 1.0.0!
Thank You¶
Thank you for being an early adopter of VAST Orbit. Your feedback shapes the future of data science on VAST DataBase.
Get Involved:
Report issues on GitHub
Join discussions on Slack
Share your use cases
Contribute ideas for new features
Happy analyzing with VAST!