Python Primers: Python for Data Science and Artificial Intelligence
Python Primers:
Python for Data Science and Artificial Intelligence
Date: Tuesday 7th April 2020
Live and Online: 09.00 – 17.00
Advanced Python Techniques
Date: Tuesday 14th April 2020
Live and Online: 09.00 – 17.00
Python for Data Science and Artificial Intelligence
Tuesday 7th April 2020: 09.00 – 17.00
Overview
Python is the de factolingua franca of data science, machine learning, and artificial intelligence. Familiarity with Python is a must for modern data scientists.
The MLI Python Primers are designed to take you from the very foundations to state-of-the-art use of modern Python libraries.
You will learn the fundamentals of the Python programming language, play with Jupyter notebooks, proceed to advanced Python language features, learn to use distributed task queues (Celery), learn to work with data using NumPy, SciPy, Matplotlib, and Pandas, examine state-of-the-art machine learning libraries (Scikit-Learn, Keras, TensorFlow, and Theano), and complete a realistic, real-life data science lab.
Syllabus:
- The fundamentals of the Python programming language and Jupyter notebooks
- Jupyter notebooks
- The Python syntax
- Data types, duck typing
- Data structures: lists, sets, and dictionaries
- Data types
- Advanced Python features; distributed tasks queues with Celery
- List comprehensions
- Lambdas
- Objects
- The Global Interpreter Lock (GIL)
- Multithreading and multiprocessing
- Distributed task queues with Celery
- Python libraries for working with data: NumPy, SciPy, Matplotlib, and Pandas
- Multidimensional arrays in NumPy
- Linear algebra and optimisation with SciPy
- Data visualisation in Matplotlib
- Time series data
- Dealing with Pandas DataFrames
- Machine Learning with Scikit-Learn; Deep Learning with Keras, TensorFlow, and Theano
- Overview of machine learning
- Introduction to Scikit-Learn
- Keras and TensorFlow
- Introduction to Theano
Tuesday 14th April 2020: Advanced Python Techniques
09.00 – 12.30: Advanced Python Features and Putting them to use in Practice
- Algorithmics and graph theory
- Prime numbers
- Cryptography
- Blockchain
- Distributed Computing with Python
13.30 – 17.00: High Performance Python
Outline syllabus
· Profiling
· The use of NumPy and SciPy over pure python
· The importance of optimised NumPy and SciPy
· The use of Cython and ctypes to integrate compiled code
· Just In Time Compilation using Numba
· Distributed computing frameworks
· Numerical precision and speed
· Using specialised instead of generalised algorithms
· NAG Library for Python
Abstract
Python is a superb prototyping language that allows us to develop high quality data analyses and simulations in a relatively short amount of time. The cost we pay for this is performance. Python is essentially an interpreted and single threaded language which puts severe limitations on its speed. In this session we will learn a range of techniques that will allow us to discover which parts of our Python code are slow and what we can do to speed things up.
Paul Bilokon:
Founder, CEO, Thalesians & Senior Quantitative Consultant, BNP Paribas
Mike Croucher:
Technical Evangelist, NAG (Numerical Algorithms Group)
MLI Online Resource Library
Prior to the start of the MLI students are given access to our online resource in preparation for the certificate.
MLI Online Primer Resource:
- Python Primer for Data Science. Presented by Nikolaos Aletras, Lecturer at The University of Sheffield
- Python for Data Science and Artificial Intelligence Workshop. Presented by Paul Bilokon: Founder, CEO, Thalesians & Senior Quantitative Consultant, BNP Paribas
- Maths Primer Refresher Material: For each topic (Linear Algebra, Optimization, Probability & Statistics), there will be a specific quiz to test initial background knowledge. We recommend that you first attempt answering the quiz exercises without looking at any material.
Additional MLI Learning Resource:
- Big Data, High-Frequency Data, and Machine Learning with kdb+/q Workshop. Presented by Paul Bilokon: Founder, CEO, Thalesians & Senior Quantitative Consultant, BNP Paribas.
Level 1: Machine Learning Institute Certificate in Finance
Level 1: Machine Learning Institute Certificate in Finance
Dates:
- Level 1 Starts: Tuesday 21st April 2020
ASSIGNMENTS:
Throughout the programme, candidates work on hands-on assignments designed to illustrate the algorithms studied and to experience first hand the practical challenges involved in the design and successful implementation of machine learning models. The data sets and problems are selected to be representative of the applications encountered in finance.
Module 1 – Supervised Learning:
In this module, the concepts related to algorithmically learning from data are introduced. The candidates are given an early taste of a supervised machine learning application before going through the fundamental building blocks starting from linear regression and classification models to kernels and the theory underpinning support vector machines and then to the powerful techniques of ensemble learning.
Module 1 Faculty:
- Adriano Soares Koshiyama: The Alan Turing Institute
The module includes a combination of theoretical and hands-on assignments:
| Module 1 | Welcome to The MLI. Supervised Learning | Theory: Learning from Data and Linear Models | 21-Apr-20 |
| Module 1 | Supervised Learning | Practical: Learning from Data and Linear Models | 28-April-20 |
| Module 1 | Supervised Learning | Theory: Ensemble Models | 5-May-20 |
| Module 1 | Supervised Learning | Practical: Ensemble Models | 12-May-20 |
| Module 1 | Supervised Learning | Theory: Kernel Methods | 19-May-20 |
| Module 1 | Supervised Learning | Practical: Kernel Methods | 26-May-20 |
End of Module 1 Assignment.
Module 2 – Unsupervised Learning:
An important and challenging type of machine learning problems in finance is learning in the absence of ‘supervision’, or without labelled examples.
In this module, we first introduce the theoretical framework of hidden variable models. This family of models is then used to explore the two important areas of dimensionality reduction and
clustering algorithms.
Module 2 Faculty:
- Paul Bilokon: Founder, CEO, Thalesians & Senior Quantitative Consultant, BNP Paribas
- Ivan Zhdankin: Associate, Quantitative Analyst, JP Morgan Chase & Co
There are theoretical and applied assignments with financial data sets.
| Module 2 | Unsupervised Learning | Introduction & Dimensionality Reduction | 2-Jun-20 |
| Module 2 | Unsupervised Learning | Practical Lab Session | 9-Jun-20 |
| Module 2 | Unsupervised Learning | Clustering Algorithms | 16-Jun-20 |
| Module 2 | Unsupervised Learning | Applications & Practical Lab Session | 23-Jun-20 |
End of Module 2 Assignment.
Module 3 – Practitioners Approach to ML:
This module focuses on the practical challenges faced when deploying machine learning models within a real life context.
Each session in this module covers a specific practical problem and provides the candidates with guidance and insight about the way to approach the various steps within the model development cycle, from data collection and examination to model testing and validation and results interpretation and communication.
Module 3 Faculty:
- Paul Bilokon: Founder, CEO, Thalesians & Senior Quantitative Consultant, BNP Paribas
- Ivan Zhdankin: Associate, Quantitative Analyst, JP Morgan Chase & Co
- Mike Croucher: Technical Evangelist, NAG (Numerical Algorithms Group)
| Module 3 | Practitioner’s Approach | Reproducibility and Deployment of Data Science Workflows | 30-Jun-20 |
| Module 3 | Practitioner’s Approach | Feature engineering / Model tuning | 07-Jul-20 |
| Module 3 | Practitioner’s Approach | Introduction to Natural Language Processing and Practical Lab Session | 14-Jul-20 |
| Module 3 | Practitioner’s Approach | Using Natural Language Processing to Predict Bond Prices | 21-Jul-20 |
End of Module 3 Assignment.
Level II: Machine Learning Institute Certificate in Finance
Level 2: Machine Learning Institute Certificate in Finance
Dates:
- Level 2 Starts: Tuesday 28th July 2020
Module 4 – Neural Networks:
Neural Network models are an important building block to many of the latest impressive machine learning applications on an industrial scale.
This module aims to develop a solid understanding of the algorithms and importantly, an appreciation for the main challenges faced in training them. The module starts with the perceptron model, introduces the key technique of backpropagation before exploring the various regularisation and optimisation routines. More advanced concepts are then covered in relation to the next module on Deep Learning.
Although we cover the theoretical foundations of Neural Networks, the emphasis of the assignments will be on hands-on lab work where the candidates are given the opportunity to experiment with the techniques studied on financial and non-financial data sets.
Module 4 Faculty:
- Terry Benzschawel: Founder and Principal, Benzschawel Scientific, LLC
- Alexei Kondratyev: Managing Director, Head of Data Analytics, Standard Chartered Bank
| Module 4 | Neural Networks | Perceptron Model | 28-Jul-20 |
| Module 4 | Neural Networks | Backpropagation | 04-Aug-20 |
| Module 4 | Neural Networks | Regularisation and Optimisation | 01-Sep-20 |
| Module 4 | Neural Networks | Network Architectures | 8-Sep-20 |
End of Module 4 Assignment.
Module 5 – Deep Learning:
Deep Learning has been the driving engine behind many of the recent impressive improvements in the state of the art performance in large scale industrial machine learning applications.
This module can be viewed as a natural follow-up from the previous module on Neural Networks. First, the background and motivations for transitioning from traditional networks to deeper architectures are explored. Then the module covers the deep feedforward architecture, regularisation for deep nets, advanced optimisation strategies and the CNN Architecture.
The assignments of this module will be highly practical with ample opportunity to experiment on financial and non-financial data sets and become familiar with the latest open-source deep learning frameworks and tools.
Module 5 Faculty:
- Blanka Horvath: Assistant Professor, Financial Mathematics King’s College London
- Ivan Zhdankin: Associate, Quantitative Analyst, JP Morgan Chase & Co
- Terry Benzschawel: Founder and Principal, Benzschawel Scientific, LLC
| Module 5 | Deep Learning | Motivation and Examples | 15-Sep-20 |
| Module 5 | Deep Learning | Reinforcement Learning: introduction | 22-Sep-20 |
| Module 5 | Deep Learning | Reinforcement Learning: implementation | 29-Sep-20 |
| Module 5 | Deep Learning | Deep Learning Volatility / Practical Lab Session | 06-Oct-20 |
End of Module 5 Assignment.
Module 6 – Time Series:
Time series data is an invaluable source of information used for future strategy and planning operations everywhere from finance to education and healthcare. You will be walked through the core steps of building, training, and deploying your time series forecasting models. You’ll build a theoretical foundation as you cover the essential aspects of time series representations, modeling, and forecasting before diving into the classical methods for forecasting time series data.
Module 6 Faculty:
- Francesca Lazzeri: Machine Learning Scientist, Microsoft
| Module 6 | Time Series | Financial Time Series Data | 13-Oct-20 |
| Module 6 | Time Series | Time Series Analysis | 20-Oct-20 |
| Module 6 | Time Series | Practical Lab Session | 27-Oct-20 |
End of Module 6 Quizzes.
Examination & Final Project
FINAL EXAMINATION:
Examination preparation lecture: Tuesday 3rd November 2020
Examination date: Tuesday 17th November 2020
- Candidates will sit a formal examination on a computer. The exam is taken online by students globally.
FINAL PROJECT:
DATE: Friday 4th December 2020
At the end of the programme, candidates apply the theoretical and practical skills acquired to a real world application within the financial services industry.
The assessment will take into account the quality and the originality of the work as well as the clarity of its presentation.
Download Event PDF