Introduction to Machine Learning

In this lecture I outline the content that will be discussed in this section and explain why data pipelines are of utmost importance in machine learning projects.

In this lecture, I discuss the data collection step and what should be considered when planning to collect data for a machine learning dataset.

In this lecture, I discuss a few techniques that are useful for conducting an explorative analysis and how one should interpret the results.

In this lecture, I discuss some of the data quality problems that can arise when constructing a machine learning dataset and how one should go about addressing them.

In this lecture, I discuss the remaining data preparation steps that should be considered when finalizing our dataset for model ingestion.

This lecture introduces two popular dimensionality reduction techniques, providing an overview regarding why they work and why one should consider applying them to high-dimensional datasets.

This lecture briefly introduces the concept of feature engineering and discusses why it is beneficial to make use of domain knowledge and exploratory analysis findings to construct additional features from the available data.

In this lecture I will introduce the field of machine learning and discuss why you should care about it as well as some of the applications for which it has already been shown to be quite effective.

In this lecture I introduce the machine learning taxonomy. I explain what the differences are between supervised, unsupervised, semi-supervised and self-supervised learning. I also discuss how conventional machine learning differs from deep learning.

In this lecture I introduce regression models, which are one of the simplest types of supervised learning models, and discuss what one would need to consider when applying them to problems.

This lecture introduces support vector machines, which are a continuation of regression models and aim to find optimal decision boundaries.

This lecture introduces the K-nearest neighbours models, which use the most similar samples from their memory to assign labels to new unseen samples.

This lecture introduces the concept of decision trees and discusses how they are used to create piecewise-linear decision boundaries.

This lecture introduces the concept of ensembles and discusses two popular tree-based ensembles, namely random forests and gradient boosted trees.

This lecture introduces the section and discusses what K-Means clustering is.

This lecture introduces the concept of hierarchical clusters and explains how divisive and agglomerative hierarchical clustering works.

This lecture introduces probabilistic clustering models, specifically Gaussian Mixture Models, which fits Gaussian clouds to clusters to determine what the probability is of each sample belonging to a specific cluster.

This lecture introduces the DBSCAN clustering algorithm and explains how it clusters samples by utilizing information regarding the density of samples in a given region.

In this lecture, I introduce the field of deep learning and provide a brief overview of what will be discussed in this section.

This lecture presents the perceptron and serves as a brief refresher regarding how linear regressors work.

This lecture explains how perceptrons can be combined to create layers and how these layers can be stacked on top of one another to form artificial neural networks.

This lecture introduces the concept of convolutional neural networks, which specialize in handling image data.

This lecture introduces recurrent neural networks, which specialize in handling sequential datasets.

This lecture introduces autoencoders and explains what they are commonly used for.

This lecture demonstrates some of the techniques that we have explained the visualize and understand datasets. Examples are provided for tabular and image datasets, with some additional content regarding text datasets.

This lecture shows how to combine some of the elements that have been discussed in this course to prepare and model a breast cancer recurrence dataset. The approach is kept relatively simple to serve as a first example regarding how to approach a machine learning project.

This lecture demonstrates what it would look like to develop a regression model that aims to predict the estimated time of arrival of vessels. This demonstration goes into more detail regarding how we would test and compare various models, as well as how one would approach hyperparameter tuning.