2
$\begingroup$

When I have finished building my machine learning model and want to make it publicly available for other users then what could be the best solution for that? I mean I'm looking for a solution where there is a UI interface for an input/output and the ML code get executed through this UI.

  1. Simple Jupyter Notebook - The code is not executable.
  2. Jupyter Notebook + MyBinder + ipywidgets - nice UI can be build by the ipywidgets and the ML code is executable but still the user needs to execute all the cell boxes before he/she can use the ML model.
  3. plotly/Dash - So far I have no experience by this. It is said that it can be used for building web applications with highly custom UI's, deploying apps to servers and then share them through URLs. Is it so that when using this then I need to have my own server where the ML code runs?
  4. Is there any solution where I don't need to have my own server and can deploy and run my ML project?
$\endgroup$
1
  • $\begingroup$ What are you looking for - Cloud hosting, Model as a Service via. API e.g. REST Or Free of cost. and Is this of Learning purpose(I am assuming)? $\endgroup$
    – 10xAI
    Commented Jul 19, 2020 at 11:57

2 Answers 2

2
$\begingroup$

The easier solution in my opinion consist to build your application using a framework like Streamlit and deploy it using docker.

Docker let you encapsulate your code and all needed package in a container, which you could think as an virtual machine (but its not); it will run on all OS since it use your configuration.

Of course there is plenty of other way but its my personnal favorite.

$\endgroup$
1
$\begingroup$

Streamlit

Streamlit is an open-source app framework for Machine Learning and Data Science teams. Create/Build an app in a few lines of code with Streamlit magically simple API. Then see it automatically update as you iteratively save the source file.

Streamlit Cloud

Streamlit Cloud is a workspace for your team to deploy, share, and collaborate on your Streamlit apps. With Streamlit Cloud your data team can directly serve the needs of the rest of the company. Quickly go from data to app, from prototype to production. Share apps in one click and collaborate instantly with live code updates.

With Community Free tier You can now deploy one private app and unlimited public apps!

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.