Data Science and AI/ML (Python) training course

Python Core Concepts and Best Practices

 

• Introduction to Python basic concepts, data structures and control flow structures. Overview of how Python is used for Data Science and Data Analytics projects.
• Environment set-up using Anaconda, a free and enterprise-ready distribution of Python. We'll discuss how to set up virtual environments and install Python packages. We'll also set up Jupyter, a web-based interactive environment where users can organise, write and run their Python code in notebooks.
• Notions of Object-Oriented Programming and Functional Programming, applied to the design of Python applications and analysis pipelines using best practices.
• The Anaconda distribution as Python Data Science platform
• Overview on Python virtual environment set-up
• Running code in Jupyter notebook
• Core data types in Python
• Control flow statements
• Defining and using custom functions
• The Python standard library

Working with data:

• Iteration and list comprehensions
• Accessing raw data on file (CSV, JSON, ...)
• Working with dates and times
• Basics of Object-Oriented Programming in Python

Python Data Science Tools

 

• We'll explore the most important Python tools for Data Science.
• NumPy, short for Numerical Python, is one of the main building blocks for scientific computing in Python. It provides high speed manipulation of multi-dimensional arrays and it's used by higher level libraries (like pandas) to support sophisticated analytics with high speed computation.
• Pandas is a highly performant library for data manipulation and data analysis in Python. It's built on top of NumPy and optimised for performance, while offering a high-level interface. We'll discuss how to create and manipulate Series and DataFrame objects in pandas, accessing data from multiple sources, cleaning and transforming data sets to get them in the right shape for advanced analysis.

Numpy:

• Working with NumPy arrays
• Essential operations with NumPy arrays
• Stats and linear algebra with NumPy

Pandas:

• Working with table-like data in pandas
• Essential operations with Series and DataFrame object
• Loading data from file into DataFrame objects
• Summary statistics over DataFrame objects
• Data aggregation queries (groupby() method)
• Exploratory analysis of new datasets
• Data visualisation over DataFrames
• Join/merge operations with DataFrames
• Working with text data in DataFrames

Databases:

• Working with relational databases in Python
• Overview on SQLAlchemy for database interaction
• Integration of pandas and SQL

Accessing & Preparing Data

 

• Data can come in multiple formats and from multiple sources. We'll examine how to read and write data from local files in different formats, and how to access data from remote source.
• Data cleaning and data preparation are the first steps in a data analysis project, so we'll discuss how to perform data transformation to get ready for further analysis.

Data Analysis

 

• With our data in the right shape, we're ready to analyse them in order to extract useful insights.
• We'll perform the computation of summary information and basic statistics from data sets. We'll approach split-apply-combine operations with Data Frames, in order to perform advanced transformations and reshaping our data with pandas.
• We'll query our Data Frames using the powerful group-by method.

 

 Data Visualisation

 

• Data analysis benefits from the visualisation of data. If a picture if worth a thousand words, complex data structures can be easier to understand and analyse using effective visualisation techniques. Communicating the results with non-technical users is also a challenge that visualisation techniques help to overcome. We'll showcase how to easily produce beautiful visualisations with matplotlib.

Miscellanea

• Python packaging: using and creating custom libraries
• Unit testing: tools to perform unit testing in Python
• Interaction with web services

Artificial Intelligence, Machine Learning and Data Science

 

• What is AI? What is ML? What's up with the hype? We'll discuss Machine Learning problems and applications. How to translate a business problem into a Machine Learning task? We'll discuss the overall process to solve these tasks and then introduce specific algorithms implemented using scikit-learn, the core Machine Learning library for Python.
• Machine Learning and Predictive Analytics
• Machine Learning is a field of computer science that gives computers the ability to learn without being explicitly programmed. Machine Learning algorithms can learn from, and make predictions on data.
• With the wealth of data available today, companies can take advantage of Machine
• Learning techniques to gain actionable insights and ultimately improve their business. Using scikit-learn, the core Machine Learning library for Python, attendees will learn how to implement Machine Learning systems to perform predictions on their data.
• We will also examine and use tools and frameworks such as the Open Source TensorFlow.

Data Exploration and Pre-processing

• Data sets come in all sorts of formats and flavours. The first part of a Machine Learning project is understanding the data and the problem at hand.
• Data cleaning, data transformation, and in general data pre-processing are the steps to perform in order to get the data sets in the right shape, so Machine Learning algorithms can learn from them.
• Python makes data exploration and pre-processing relatively easy.

• Feature Engineering

• By injecting domain knowledge in the process, attendees will learn how to extract attributes from the data and how to encode them into features that make Machine Learning algorithms work.
• One of the core aspects of applied Machine Learning, feature engineering is difficult and time consuming.
• The quality and quantity of features can have a great impact on how Machine Learning algorithms can work.

SUPERVISED LEARNING

• In supervised learning, the training data consist of a set of training samples associated with a desired output label. Supervised learning algorithms can learn the desired output from the training data, and make a prediction on new, unseen data.
• We'll approach supervised learning from two different directions: classification, the task of predicting a category, and regression, the task of predicting a quantity.
• Examples of applications include price prediction, spam detection and sentiment analysis.

UNSUPERVISED LEARNING

• In unsupervised learning, the training data is not labelled. Unsupervised learning algorithms analyse the data and find hidden structures within the data.
• We'll approach unsupervised learning in particular from the point of view of a clustering application.
• Examples of applications include social network analysis, customer segmentation or product recommendation.

MACHINE LEARNING EVALUATION

• Using the proper evaluation metrics, we can understand how well our algorithms are performing and we can compare the performances of different algorithms.
• Attendees will learn about error analysis and model introspection, in order to “debug” and improve Machine Learning algorithms

Neural Networks and Deep Learning

 

• A high level introduction to Artificial Neural Networks, a family of algorithms applicable to many Machine Learning problems, and relevant mathematical concepts. Discussion on Neural Network concepts like gradient descend, activation functions, loss functions and hyper-parameters.