FA-0791Data & AnalyticsSoftware Development

Machine Learning Roadmap with Python

Three phases from programming basics to guided models and evaluation

Introduction

Why this course

Progress through Python foundations, data analysis and introductory machine learning in three sequential phases. Use prepared examples to understand modelling, predictions and evaluation. The final three-day phase is a broad guided survey: selected algorithms receive hands-on practice and others are demonstrations, not independent mastery of every method.

Learning outcomes

Translate simple problems into algorithms and Python code.
Use functions, modules, collections, exceptions and a notebook/editor environment.
Acquire, clean, manipulate and visualise data with NumPy, pandas, Matplotlib and Seaborn.
Compare supervised, unsupervised, recommendation, bandit and neural-network examples.
Prepare data without leaking held-out information and evaluate a guided project.

Prerequisites

Ability to use a computer
Ability to use the internet
Manage files and understand basic file systems in the computer
Ability to communicate in the English language

A suitable computer, browser, internet and access to the prepared compatible Python/library environment. No prior Python knowledge is required for Phase I; later phases build on earlier practice.

Training outline

6 modules

01Phase I — Python development foundations (3 days)1 topics

Programming concepts

What Is Algorithm
What Is Programming
What Is Program
Machine Language
Programming Language
What Is Translator

Python introduction

Brief History of Python
Python Versions
Installing Python
IDEs
Environment Variables
Python Documentation
Hello world with python
Modes of Programming

Google Colab notebooks

What Is Colab Notebook
- Why Colab Notebook Is Important
- Accessing hosted Colab and selecting a runtime
- Main Components of Colab Notebook
- Modes
- Exporting Notebook Documents
Basic Python syntax
- Identifiers
- Reserved Words
- Lines and Indentation
- Comments

Variables and collections

Numbers
Python Lists
Python Tuples
Python Dictionaries
Python Sets
Copying
Python Strings
String formatting
Regular Expressions

Python operators

Arithmetic Operators
Comparison (Relational) Operators
Assignment Operators
Logical Operators
Membership Operators
Operators Precedence

Control flow

Decision Making
The if Statement
The if else Statement
For Loop
While Loop
Break And Continue

Functions

Defining Your Own Functions
Parameters
Function Documentation
Passing Collections to a Function
Variable Number of Arguments
Scope
Map
Filter
Lambda

Modules and packages

What Are Modules
Importing Modules
Aliasing
Importing Set Of Element From A Module
Namespace
What Are Packages
dir function
help function

Exceptions

What Is Exception
Exception Types
Try Except Component
Handling General Exception
Handling Specific Exception
Raise

02Phase II — Data workflows and analytics (2 days)5 topics

External-library use in notebook, desktop, server and remote contexts.
NumPy and pandas for acquisition, pipelining, ingestion, cleaning, conformance and manipulation.
Files, encoding, extraction and conversion.
Matplotlib and Seaborn for visual and preliminary analysis.
Introduction to prediction problems in preparation for Phase III.

03Phase III — ML process and preprocessing (shared 3-day phase)3 topics

Recursion
Memoization
Selected advanced library concepts

Fundamentals of Machine Learning

What is Machine Learning?
Process of Machine Learning
Life Cycle of Machine Learning
Application working in Machine Learning
Types Of Machine Learning
Setting up the development environment lab

Data Preprocessing

Importing Libraries
Importing Dataset
Taking care of Missing Data
Encoding Data: Categorical Data
Splitting the dataset into the Training set and Test set
Feature Scaling

Split training/test data before fitting imputation, encoding or scaling; learn transformations from training data and apply them consistently to held-out data.

04Phase III — Supervised learning examples2 topics

What is Supervised Learning?
Types of Supervised Learning

Regression

Simple Linear Regression
Multiple Regression
Polynomial Regression

Decision Tree

Implementation of Decision Tree
Random Forest Regression
- Implementation of Random Forest Regression
Classification
Implementation of Logistic Regression
- Implementation of Naive Bayes
- Implementation of Support Vector Machine (SVM)

Introduce XGBoost as gradient boosting for supervised learning, not as a neural-network architecture.

05Phase III — Clustering, associations and recommendations4 topics

K-means Clustering
K-means Selecting the Number of Clusters
Implementation of k-means clustering
Implementation of Hierarchical Clustering

Apriori association-rule mining is separate from clustering.

User based
Item based
Implementation

06Phase III — Bandits, neural networks and NLP1 topics

Introductory bandit examples

Implementation of Upper Confidence Bound (UCB)
Implementation of Thompson Sampling

Neural-network demonstrations

Implementation of Artificial Neural Networks (ANN)
Implementation of Convolutional Neural Network (CNN)

Dimensionality reduction

Implementation of Principal Component Analysis
Implementation of Linear Discriminant Analysis
Implementation of Kernel PCA

Linear discriminant analysis uses class labels for supervised reduction, unlike PCA.

NLP and project

Implementation of Natural Language Processing in Python

Hands on Project

Evaluation

Discuss potential business uses and the relationship among AI, ML and deep learning; evaluate examples rather than guaranteeing accurate future predictions.

A programme built around your team.

Share your training goals and requirements.