(image: xkcd)

Topics

• Course setup and logistics
• Machine learning vs. deep learning, AI, statistical analysis
• Workflow and tools
• Setup the environment

Logistics

Discussions, questions: mldds03.slack.com

Structure: 5 modules

Assignments: two graded projects and presentations

1. Practice project at the end of module 2
2. Final project due by the end of this course (module 5)

What to expect

Python coding is required

Lots of experimentation

Practice

Patience

What this is not

High level strategy

Solving math equations

Implementing altogithms from scratch

Exhaustive coverage of all algorithms

We'll program machines to ...

Label things [e.g. Classification]

Predict trends [e.g. Linear prediction]

Find groups of things [i.e. Clustering]

Find outliers [i.e. Anomaly detection]

... maybe more

Activity: what's the difference between ... ?

Data Science Process in 1 Slide

(image: szilard)

Machine Learning Tools

Informal survey: https://github.com/szilard/kaggle-scripts-R-pydata

Tools we will use

Programming language: Python 3

Environment: Jupyter and Anaconda

Libraries: IPython, numpy, pandas, matplotlib, scikit-learn, keras, nltk, DeepSpeech, ...

Experiment! Don't limit yourself only to the tools we cover

Data Science Processes & Tools

Data Collection: not covered because data-source dependent (HTML, XML, JSON, databases, images, audio, video, ...)

Data Visualization: numpy, pandas, matplotlib

Data Transforming: numpy, pandas

Model: scikit-learn, keras, etc

Validation: scikit-learn, keras, etc

Workshop: Setup Environment

Install Anaconda for Python 3

Go to https://www.anaconda.com/download/

Download and install Anaconda for Python 3.6

Open the cheatsheet: https://conda.io/docs/_downloads/conda-cheatsheet.pdf

Find conda

Get a command prompt with conda in your path:

Windows: Start Button -> "Anaconda Prompt"

Ubuntu / MacOS: conda should be in your path

Create environment

Locate the command in the cheatsheet on "create a new environment ..."

Something that starts with conda create ...

Create an environment called mldds01 for Python 3:

conda create -n mldds01 python=3

Use environment

Activate the environment, following the output from conda create.

You can also find this from the cheatsheet.

Install packages

We'll be using Jupyter notebooks as our IDE (interactive development environment) for machine learning experiments.

In your brand new conda environment, install Jupyter:

conda install jupyter

Start Jupyter

Browse your command line to the location of this notebook (you can use cd)

cd path/to/mldds-courseware
jupyter notebook

In the browser window, navigate to 01_GettingStarted, then the first workshop: numpy.ipynb.