Skip to content

README.md

Latest commit

 

History

History
64 lines (39 loc) · 3.26 KB

README.md

File metadata and controls

64 lines (39 loc) · 3.26 KB
Raw

Project name is currently a placeholder, and will be changed later

_datascience-tool is a tool for simplifying common computing tasks data scientists and ML engineers perform on datasets. These include tasks such as combining datasets, creating row-wise aggregations, visualizing & exploring data. This is mostly dull-to-write and repetitive code that interfaces with pandas and matplotlib. This routine data manipulation and visualization code is repetitive and tedious to write, and _datascience-tool aims to automate these common data manipulation tasks.

So to summarize the project in one sentence: automation to routine data visualization, combining, aggeration, and column creation tasks using Pandas and Matplotlib, all managed through a clean user interface.

Setup

  • Packaging Python runtime and projects Python files and dependencies to a binary using pyoxidizer
  • Embedding that Python binary into a tauri desktop application
  • Running the Python binary (FastAPI web server) from within Tauri's rust backend
  • Now the frontend client (Chromium based) can call the Python server running locally without being dependent on or interfering with the system's Python.
  • This allows for creating a GUI for the common data manipulation and visualization tasks done in the Python ecosystem while making the application easy-to-use and well-contained.

Currently supported features (project is still being developed)

  • Import & export datasets
  • Quickly inspect the rows of very large datasets using pagination (tested to be quick on +20m rows)
  • Fine-grained filtering options based on the column type (categorical, date, number etc.)
  • Combine datasets, create row-wise aggregations, and other routine data manipulation
  • Create new datasets based on the existing ones
  • Cross platform (tested to work on Mac and Windows, Linux will come soon)
  • Visualize data
image

Dev requirements

  • Python
  • Rust
  • Node (v18 or higher)

Env vars

DEV_PYTHON_PATH Add this to .env within the src-tauri directory. This should point to the Python runtime that you want to use for development.

Run locally

Before running the following steps, you should make sure that the Python runtime has the required dependencies. The full list of dependencies can be found at: src-server/requirements.txt.

  • git clone https://github.com/JNeuvonen/_datascience-tool
  • cd _datascience-tool
  • pip install -r src-server/requirements.txt
  • python scripts/build.py
  • npm install
  • npm run tauri dev

Building for prod

There's an additional requirement of installing pyoxidizer for packaging the python runtime into a binary.

  • Run python scripts/build.py. This creates the python runtime binary that is used in production.
  • Set PROD_PYTHON__BUILD env var within src-tauri to point to the pyserver executable that the previous step created, here's an example: PROD_PYTHON_BUILD="/Users/Jarno/Documents/_datascience-tool/src-tauri/binaries/build/x86_64-pc-windows-msvc/debug/install/pyserver"
  • Run npm run tauri build
  • Artifacts will be found from src-tauri/target