LiQuer Framework
Web Framework for Data Science
LiQuer is a leightweighted open-source framework for managing data-science projects
from exploration phase to end-user applications.
LiQuer is designed to be easy to use, easy to learn, easy to extend and easy to integrate with other frameworks.
LiQuer puts emphasis on the transparency, traceability and discoverability of the data-science projects.
LiQuer is a versatile tool - it helps to create interactive dashboards and web applications, working with tables, creating charts, images, reports.
LiQuer can be used to build both interactive applications and non-interactive batch processes.
At the core of Liquer is a minimalistic query language (Link Query), representing a pipeline, i.e. a sequence of actions.
In its compact form it can be used as a part of URL "link" - hence the name Link Query.
This query language is human readable and easy to learn. It is useful in multiple contexts:
- as a pipeline definition for data-science projects,
- as a rich interactive REST API
- in batch process definition
- as a link identifying data sets, documents and other resources
- as unique key for caching and long term storage
LiQuer Features
Simplicity, Flexibility, Transparency and Interoperability
Friendly
LiQuer is simple to use. Just create a python function and add a decorator to it.
Web-Powered
LiQuer allows to access all its data and functionality via web interface.
It empowers data scientists to create and host interactive dashboards, universal analytical tools and complete web applications.
Document-oriented
LiQuer supports automatically generated interactive and non-interactive documents and reports,
while maintaining the reproducability and tracking the origin of the data in the metadata.
Each document and data asset can be identified and re-created by its LiQuer link.
Rich Metadata
Every asset created in LiQuer is equipped with metadata, describing its origin, processing history and other properties.
This metadata is stored in a structured way and can be used for searching, filtering and other operations.
Metadata are suitable for data-scientists e.g. for cataloguing the experiments,
but as well for auditing purposes in regulated environments - e.g. for tracking data lineage.
Modular
LiQuer is designed to be modular. It is technology-agnostic and can be used with any data-science stack.
For example: LiQuer supports multiple data-frame libraries, multiple charting libraries and multiple web frameworks as backends.
LiQuer can be extended in multiple ways: by adding new data-frame libraries, new charting libraries, new web frameworks, new data sources, caching and storage options, new data sinks, new interactive frontend tools.
Link Query
Link Query is a minimalistic query language, living in a sweet spot between simplicity and expressiveness.
It is human readable and easy to learn. It is useful in many contexts.
A huge benefit is its ability to represent the whole pipeline leading to creation of any data asset, chart or document.