Hands-on Tutorials

Jupyter notebooks have a somewhat poor reputation in the wider programming community. Joel Grus’ famous “I don’t like notebooks” talk, which he bravely gave at JupyterCon in 2018, covered many of the reasons why. Typically notebooks are seen to promote poor coding practices as they can be difficult to version control, often rely on cells being run in a specific order to return correct results and can be tricky to test and debug.

However, Jupyter notebooks are still the number one choice for most data scientists when it comes to performing tasks such…

Natural language processing, or NLP combines the fields of linguistics, computer science and artificial intelligence to enable machines to extract meaning, insights and make predictions from text data.

The majority of modern organisations collect large quantities of text data from their customers and business operations in the form of emails, chats, phone calls and many other interactions. This data can be invaluable in developing a deeper understanding of the problems that customers are facing, and insights from this data can help to solve some of these problems as well as automate and optimise customer facing business processes.

It is therefore…

Data analysis is fundamentally about finding answers to questions with data. When we perform some calculation or compute a statistic for a set of data it is usually not enough to do that across the entire dataset. Instead we will usually want to split the data into groups, perform the computation and then compare the results across different groups.

Data analysis is fundamentally about finding answers to questions with data.

Let’s say we were a digital marketing team investigating the potential reasons behind a recent decline in conversion rate. Looking at conversion rate as a whole over time would be…

Over the last few years I have written serveral articles about learning data science using online resources. During my own learning journey I have identified some of the best free or low cost material available for learning data science.

I recently spent some time consolidating this list into this Github repository so that it can be used as a quick reference for anyone who wants to expand their data science skills. …

Getting Started

Statistics is “a branch of mathematics dealing with the collection, analysis, interpretation, and presentation of masses of numerical data”. Throw programming and machine learning into the mix and you have a pretty good description of the core skills for data science.

Statistics is used in almost all aspects of data science. It is used to analyse, transform and clean data. Evaluate and optimise machine learning algorithms. It is also used in the presentation of insights and findings.

The field of statistics is extremely broad and determining what exactly you need to learn and in what order can be difficult. Additionally…

The Python programming language has many different versions. Similarly, all Python libraries also have multiple versions, work with specific versions of Python and most of them depend on other packages to run, this is known as a set of dependencies.

Every data science project that you undertake is likely to require its own unique set of third-party Python packages. Virtual environments act as self-contained environments encapsulating the Python version and all dependencies for a project. Creating a new virtual environment is one of the first steps that is usually taken when starting any new data science project.

Creating a new…

Statistics is a fundamental skill that data scientists use every day. It is the branch of mathematics that allows us to collect, describe, interpret, visualise, and make inferences about data. Data scientists will use it for data analysis, experiment design, and statistical modelling.

Statistics is also essential for machine learning. We will use statistics to understand the data prior to training a model. When we take samples of data for training and testing our models we need to employ statistical techniques to ensure fairness. …

When I started learning data science a few years ago most job ads requested a PhD, or at the very least a masters, in maths, statistics or a similar subject as an essential requirement.

Over the last couple of years, things have evolved. With the development of machine learning libraries that abstract away much of the complexity behind the algorithms, and a realisation that practically applying machine learning to solve business problems requires a set of skills that are not usually acquired through academic study alone. …