Skip to main content
Back to Blog

What is machine learning in data science? A beginner's guide for New Zealand

Updated on November 19, 20266 minutes read


Picture a Wellington power retailer trying to guess which customers are about to switch providers next month. A human analyst could eyeball a few hundred accounts and make a decent call. Machine learning in data science lets you make that same call across half a million accounts, update it every night, and improve it as new data arrives. That gap — between hand-checking and learning from data at scale — is the whole point.

If you've asked whether data science and machine learning are the same thing, the short answer is no. Machine learning is one tool inside the much bigger job of data science. This guide breaks down what that tool does, where it fits, and how a beginner in New Zealand can start using it.

What machine learning actually means

Machine learning is a way of writing programs that find patterns in data instead of being told every rule by hand. You don't code "if the customer hasn't logged in for 60 days, flag them". Instead you feed the model thousands of past examples — customers who left and customers who stayed — and it works out which signals matter and how much.

Here's a concrete example a total beginner can picture. Say you want to predict house prices in Christchurch. You give the model a spreadsheet: each row is a past sale, with columns for floor area, number of bedrooms, suburb, and section size, plus the price it sold for. The model studies those rows and builds a formula that maps features to price. Show it a new listing it has never seen, and it estimates a price. Feed it more recent sales and it adjusts. That adjusting part — getting better from more examples — is the "learning".

Not every problem needs this. If you just want to count how many sales happened last quarter, that's a SQL query, not machine learning. ML earns its place when the pattern is too messy or too large for a fixed set of rules.

Where machine learning sits inside data science

Data science is the broader craft of turning raw data into decisions. A working day might include pulling data out of a database, cleaning out the junk, making charts to understand what's going on, and reporting findings to a team. Machine learning is the step where you build something predictive on top of that groundwork.

A useful way to think about the relationship: data science is the kitchen, and machine learning is one of the appliances. The appliance is powerful, but it's useless without prep, clean ingredients, and someone who knows what dish they're making. In practice, data scientists spend far more time cleaning and understanding data than training models. The model is often the fastest part.

This also answers a question people often ask about how the field connects to AI. Machine learning is the engine behind most modern artificial intelligence, so it sits in the overlap between the two. We've written more on how data science and AI fit together if you want the fuller picture.

The main types you'll meet

Most beginner machine learning falls into two camps, and it helps to know which one a problem belongs to before you touch any code.

Supervised learning uses labelled examples — data where you already know the answer. House prices, spam detection, loan approvals. You know the past outcomes, so the model learns to predict future ones.

Unsupervised learning has no labels. You're asking the model to find structure on its own, like grouping customers into segments nobody defined in advance. An Auckland e-commerce team might use it to discover that shoppers naturally cluster into a few buying patterns they can then market to differently.

There's a third area, reinforcement learning, where a system learns by trial and error against a reward — think of a model learning to control inventory or route deliveries. It's less common in entry-level data roles, so don't stress about it early on.

Supervised vs unsupervised learning at a glance

AspectSupervised learningUnsupervised learning
Needs labelled data?Yes — you know the right answersNo — the model finds its own structure
Typical question"What will this value be?""What natural groups exist here?"
Beginner examplesPrice prediction, churn, spam filteringCustomer segmentation, anomaly spotting
Common NZ use caseForecasting demand for a retailerGrouping users for a marketing campaign
How you check it worksCompare predictions to known answersHarder — relies on judgement and metrics

Tools you'll actually use

The day-to-day stack for machine learning in New Zealand data teams is mostly Python. You'll write code in Jupyter notebooks or VS Code, wrangle data with pandas, and build models with scikit-learn. For bigger neural-network work, people reach for TensorFlow or PyTorch. SQL stays essential because the data almost always lives in a database before it ever reaches a model.

You don't need all of this on day one. A beginner can get a real model running with Python, pandas, and scikit-learn alone. Everything else you add as the problem demands it.

Can you start after school or a career change?

Yes, and you don't need a maths degree to begin. Plenty of people move into data work from finance, science, teaching, or operations roles across Auckland, Wellington, and the wider country. School leavers can start too — if you can handle NCEA-level maths and you're willing to learn Python, you have enough to begin.

What matters more than your background is steady practice on real datasets. Reading about models teaches you the vocabulary; building a churn predictor or a price model teaches you the job. If you prefer a structured route with projects and feedback, a focused data science and AI bootcamp covers the full pipeline rather than a single tool. If you'd rather set your own hours around a job, the self-paced data science and AI course runs through the same ground at your pace.

A realistic first project

The fastest way to understand machine learning is to build one small thing end to end. Grab a public dataset — NZ open data portals and sites like Kaggle have plenty. Pick a clear question, such as predicting whether a flight is delayed. Load it with pandas, clean the obvious mess, split it into training and test sets, train a simple model, then check how often its predictions match reality.

You'll learn more from that one messy project than from a dozen tutorials. The cleaning will annoy you, the first model will underperform, and fixing it is where the real skill grows.

Machine learning in data science is one technique with an outsized reputation — genuinely useful, but only as good as the data prep and problem framing around it. Start small, build one model end to end, and let the gaps you hit guide what you learn next. If you want a clear path and want to see what a structured programme covers, compare your options on the Code Labs Academy course pricing page.

Learn technical skills online with Code Labs Academy

Learn technical skills online with Code Labs Academy

Join our supportive community, unlock your potential, and embark on a rewarding career path.

Apply Now

Frequently asked questions

What is machine learning in data science?

Machine learning is a technique inside data science where a program finds patterns in data instead of following hand-written rules. You feed it past examples, and it learns to predict or group new data — like estimating a house price from past sales. It's one tool in the wider data science job, which also covers collecting, cleaning, and interpreting data.

Is data science a part of artificial intelligence?

Not exactly — they overlap rather than one sitting fully inside the other. Data science is about turning data into decisions, while AI is about building systems that act intelligently. Machine learning is the shared engine between them, which is why the two fields are so closely linked but not identical.

Can I join a data science or AI course after school in New Zealand?

Yes. School leavers can start a data science and AI pathway without a university degree. If you're comfortable with NCEA-level maths and willing to learn Python, you have enough to begin. Steady practice on real datasets matters more than your formal background.

What is the difference between supervised and unsupervised learning?

Supervised learning uses labelled data where the answers are already known, such as past house prices, so the model learns to predict future values. Unsupervised learning has no labels and finds structure on its own, such as grouping customers into segments nobody defined in advance.

What tools do I need to start machine learning?

Most New Zealand data teams use Python with pandas and scikit-learn, written in Jupyter notebooks or VS Code, plus SQL for pulling data from databases. Larger neural-network work uses TensorFlow or PyTorch, but a beginner can build a real model with just Python, pandas, and scikit-learn.

Career services

Personalized career support to help you launch your tech career. Get résumé reviews, mock interviews, and industry insights, so you can showcase your new skills with confidence.

Explore