Start using craft ai in a Python Jupyter Notebook app using the official client, in a NYC taxi context.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
data
src
.editorconfig
.gitignore
.travis.yml
LICENSE
Pipfile
Pipfile.lock
README.md

README.md

craft ai Jupyter Notebook starter kit

Build Status License

craft ai cognitive automation API leverages explainable Artificial Intelligence to 10x your knowledge workers productivity. craft ai is the first high level AI API enabling Automated Machine Learning at the individual level that generates explainable predictive models on the fly.

This repository hosts a fully working notebook, in a NYC taxi trip context, integrating craft ai written with Jupyter using craft ai official Python client.

The end goal: give an insight of the best New York city areas to find clients as a taxi driver. Using craft ai to analyze the 2017 yellow taxi trip records, this simple application learns when each NYC areas needs taxis.

Setup

  • Download or clone the sources from GitHub,
  • Install Python v3.7 on your computer, alternatively install any version of Python and install pyenv,
  • Install pipenv to properly manage dependencies,
  • Install the dependencies including craft ai python client, by running pipenv install in the cloned or downloaded repository, from a terminal.
  • Create a project following the subsection 1 of this tutorial and copy the write token
  • in this directory, fill a .env file setting the following variable:
    CRAFT_TOKEN=paste-your-token-here

Run

The following command will allow the user to access three notebooks: Preprocessing.ipynb, Main.ipynb and Benchmarks.ipynb. We recommend to start by running Main.ipynb.

pipenv run notebook
  • The principal work is done in Main.ipynb.
  • Preprocessing.ipynb has been written to generate the dataset yellow.csv
  • Benchmarks.ipynb allows the user to have a insight the Craft AI's agents performances.

What do next ?

You can use this use case as a craftai.pandas example, and start your own projects with the tools provided by the library.

About the dataset

The work is based on the dataset yellow.csv located in the directory data/. (It is possible to regenerate this dataset by using the notebook Preprocessing.ipynb.)

yellow.csv has been extracted from the data available on the NYC Taxi and Limousine Commission (LTC) webpage.

The craft ai user documentation can be found at https://beta.craft.ai/doc and technical questions can be sent by email at support@craft.ai.