This project is no longer mantained, since most features intendend are offered by JuliaBox
I think it's time we blow this scene. Get everybody and the stuff together. Ok? 3, 2, 1: Let's jam. Tank, Yoko Kanno (Cowboy Bebop opening theme)
It provides capabilities for basic pipeline of analysis (e.g. exploration, visualization, diagnostics) given an ordinary Julia function (e.g. linear regression lm(@formula(y~x),data)
).
JAM.jl enables one to play with a band of models.
Version : 0.0.2
Date : 11 Jul 2019
Maintainer: Felipe C. Argolo @argolof
JAM.jl exposes functions through RESTful API and widgets for interactive management and visualization of objects. Once a functionality is loaded, it can be activated with POST requests http://jam-url.com/run/my-linear-model/
containing adequate arguments.
The following example takes a pair of vectors (X,Y) and fits a linear regression model to it.
# Data frame df is generated in /test/generate_samples.jl and contains variables x and y_linear
julia>json_load = """
{ \"source\" : \"json_stream\",
\"x\" : $(df.x),
\"y\" : $(df.y_linear)}"""
julia> url_query = "http://jam-url.com/run/my-linear-model/"
julia> HTTP.request("POST", url_query,["Content-Type" => "application/json"],json_load)
HTTP.Messages.Response:
"""
HTTP/1.1 200 OK
(...){"error": false, "result": "Outputs saved to lm-2019-07-11T16:06:07.821"}"""
File lm-2019-07-11T16:06:07.417.csv contains outputs.
Data and relevant outputs are stored and can be retrieved later for data visualization, exploration and diagnostics in http://jam-url.com/display/
.
Visiting the web page http://jam-url.com/run/my-linear-model/
or using the GET method will yield information about methods supported:
Hello! Methods for my-linear-model are: Method[lm(X, y) in GLM (..)
* Methods to import functionalities (**Beagle.jl**, **Abu.jl**)
* Import models and create adequate endpoints.
* Import widgets and add them to the display available list.
* `Display` screen.
* Sidepanel with IDs from analysis
* Standard X-Y Widget (in Beagle.jl)
* List of models/datasets screen (View of SQL DB)
* Init Docker containers for predict tasks (**Infra.jl**)
* Built-in system to keep track of dataset versions and analysis using MD5 checksum or similar.