Some notebooks used by Dato in GraphLab Create training sessions.

For your convenience, a shortened link to this repo: https://goo.gl/ov6ZOq .

For TAU students (and Linux users in general), here are the GLC Installation instructions on Linux. Python, pip and virtualenv are assumed to be installed.

For others, you can install GraphLab by pip install -U graphlab-create or follow the download and installation instructions here. Note that you'll need to register for a product key.

The Shadows of Data Science slides are available here.

The data for these notebooks consists of a recommender systems dataset which can be downloaded from these addresses. It should be placed in the same folder as the training IPython Notebooks.

• https://s3.amazonaws.com/dato-datasets/dato-training/review.csv.gz (80.4 MB)

Business reviews written by users. Each review also includes a rating (ranging from 1 to 5 stars). As such, it can be used for training a recommender system, as it describes user-item relationships (user-business in this case) along with a target variable (the stars rating).

• https://s3.amazonaws.com/dato-datasets/dato-training/business.csv.gz (640.2 KB)

Side features describing the businesses appearing in the reviews.

• https://s3.amazonaws.com/dato-datasets/dato-training/user.csv.gz (1.1 MB)

Side features describing the users who wrote the reviews.

Below are the currently available notebooks. More notebooks will be added soon!

[Basic Training](Dato Basic Training.ipynb) - tabular data maniuplation using SFrames, graph analytics using SFrame and our PageRank toolkit, and some predictive modelling using our recommenders toolkit.

[Advanced Exercise](Dato Exercise.ipynb) - an exercise consisting of re-creating the backend of our Pathyways Recommender demo. This exercise is harder than the ones given in the training, and can be considered a 'home work' for enthusiasts.

More notebooks on many different topics can be found in the notebook gallery.

More information about our toolkits can be found in our API documentations.

Users willing to learn GraphLab Create in its entirely should go through our user guide.

It took us 4 months to reach 40,000 students in our Coursera courese - Machine Learning Foundations. Have a look! The first course in this specialization can be taken for free.