MLSQL Lang example project

This is a mlsql-lang example project.

Installation Of MLSQL Lang IDE

Download Visual Studio Code
Download MLSQL Lang VSCode plugin
- Winwdows
- Mac
- Linux

After installation of Visual Studio Code , switch to Extensions tab, click ... on right side of search bar, find the item Install from VSIX..., choose the VSCode plugin we had already downloaded in preview step and intall it.

Notice that for now we only support Ligt theme in vscode. Select Code > Preferences > Color Theme :

In command palette popup，select the light color:

Download this project

Click Clone or download and select Download ZIP, then you get the package of this project. Unzip it in you desktop.

Open this project in vscode

Select File > Open... and choose the location where we unzip this project.

Tutorial

./src/try_mlsql is a good start point for you to learn MLSQL.
./src/a_tour_of_mlsql you can learn full picture of MLSQL Lang.
./src/examples/examples there are many mlsql code snippets in this notebook.
./analysis/example/cifar10/ResizeImage teach you how to processing image distributly.
./analysis/example/cifar10/DistributeTFTrainning teach you how to train DL by tensorflow distributly and deploy model as UDF .

Python Script Support

Ray is a build-in plugin in MLSQL which can execute Python script.

The power part is that you can access the data in target table in Python and pass the result processed back as a new table.

Some limitation for now:

The schema of python output should be specified mannually.

!python conf "schema=st(field(a,long))";

The basic python dependencies:

pyarrow==4.0.1
ray[default]
aiohttp==3.7.4
pandas>=1.0.5; python_version < '3.7'
pandas>=1.2.0; python_version >= '3.7'
requests
matplotlib~=3.3.4
uuid~=1.30
pyjava

Suppose you can have created virtual python enviroment called ray1.8.0 (this will used by example in this project by default).

conda create  --name ray1.8.0 python=3.6