Examples of Using Dotscience

This section will contain some tutorials showing various ways that dotscience can be used.

The tutorial material is not yet live.

In the meantime, you can

The tutorial content that will be posted here is as follows:

Quick Start

How to get up and running in dotscience, and run a machine learning model

Roadsigns Full Example

R0. Roadsigns - Introduction

Train a machine learning model to recognize road signs and deploy it into production
See various benefits of using dotscience to do this

R1. Roadsigns - Get the Data

Get the data, in this case showing how to access it from Amazon S3
Data can be accessed from anywhere your favorite tool can connect to

R2. Roadsigns - Prepare the Data

Prepare the data, showing generic tool use, and dotscience’s data provenance + data run tracking
Full data prep pipelines can be instrumented similarly

R3. Roadsigns - Train a Machine Learning Model

Train a machine learning model, in this case a simple Tensorflow/Keras neural network, to recognize road signs
Show how model tracking and provenance is automatic in dotscience

R4. Roadsigns - Deploy the Model into Production

Deploy our roadsign predictor ML model to production with Dotscience and Tensorflow serving
Show how production deployment is considerably simplified versus setting up your own from scratch

R5. Roadsigns - Model Monitoring in Production

Statistically monitor the roadsigns model after deploying it

Other Tutorials for Jupyter

These will showcase other tasks that can be performed in dotscience via the Jupyter Lab/Notebook interface

J1. DotScience & Jupyter - Interactive Development in Python with JupyterLab

Interactively develop models using Jupyter embedded in Dotscience

J2. DotScience & Jupyter - Integration with Git Version Controlling

Checking out and managing a GitHub repository inside Jupyter with Dotscience SSH key integration

Other Tutorials for the Command Line Interface

These will show how tasks can be performed directly with the dotscience command line interface, as opposed to Jupyter

C1. DotScience Command Line - ds run Interactive Development

Interactively develop models with your choice of IDE and Terminal using the ds run command

C2. DotScience Command Line - ds run Git Integration

Cloning a git repository from GitHub using the ds run command

C3. DotScience Command Line - ds run Integration with Amazon S3

Accessing a dataset stored in an Amazon S3 bucket with ds run while tracking provenance

C4. DotScience Command Line - ds run Integration with a CI System

Triggering Dotscience runs on pushes to GitHub in a continuous integration (CI) system


In future, we will add examples of further dotscience functionality, and other data science analyses.