We are fortunate to have keynote speaker, Simon Young, Head of Product Development at TradeMe (the NZ eBay equivalent for those of you on the other side of the pond), presenting at our inaugural Chief Data & Analytics Officer NZ event, 28-29 November. Simon's shared a guest post below to give you some insight into how Trade Me have been leveraging data science to create data-driven services and enhance the customer experience.

Like fidget spinners, cheap Auckland houses, and insightful yachting wisdom, Data Science is something that’s hot on Trade Me right now.

We’re always looking for ways to put our technology to work to make things easier for our members and Data Science is right up our alley. It’s a fancy way of using scientific methods to extract insights from data (often vastly hugely mind-bogglingly large amounts of data) in order to make our products better for our members.

We’ve been working with data here and there for many years and recently spooled up our first official Data Science squad. We’re a small team, but armed with a lot of data and a fair bit of science, we plan to be releasing small improvements to the Trade Me experience every few weeks.

Here’s a quick highlight of three services we’ve already released:

Suggest a category

Choosing the right category is an important step in listing an item but can sometimes be time-consuming and confusing. Luckily this is the kind of challenge that is perfect for an algorithm and a trained model.

Every day over 700,000 listings are created on Trade Me which means lots of words are being assigned to categories all the time. When it comes time to list your “iPhone 5c” or “Red t-shirt” we have a pretty good idea how to classify those words into a category and help put your listing where it belongs on Trade Me.

Thanks to the work of our Data Science squad, we can now suggest categories for Marketplace items with a high confidence (with support for motor vehicles and other kinds of listings coming soon). You can try this for yourself by starting a listing on our new Preview site. Pop in a title and we’ll recommend the top three categories we think your listing should go in.

Suggest a start price

Another challenge is knowing what price to put on a listing. With about 35,000 things sold on Trade Me each day, we have a pretty good idea what price things sell for.

When you’re creating your listing on our new Preview site, we’ll look at the title, the category, and the item’s condition, then give you a range of prices to set as your start price based on the sale prices of similar items.

It works well for things that sell frequently, but is much trickier if you’re selling something that is very rare (so we’ll leave that up to you).

Insights into property value

Property Insights is a neat service we offer to those wanting to better understand property prices. Whether you’re an agent, a home-owner, or perhaps looking to enter the property market, having an idea about how much houses are selling for is pretty handy. Data Science to the rescue!

We’ve been working with local councils all over the country and have built a sophisticated model that produces a computer-generated estimate of what a residential property might sell for based on RV data, sales information and user data on Trade Me.

We think it’s pretty great - but of course it’s just an estimate and doesn’t take into account any changes to a property that we don’t know about. If you’re wanting a better idea of what your property is worth then we recommend having a chat to a friendly property agent.

Only just begun

We’ve only just scratched the surface of what we can do with Data Science at Trade Me. Some of the things we’re working on we hope you’ll never directly notice, aside from a vague, warm feeling of things being a bit easier and a little more intuitive.

There’ll also be things we hope you will notice, like a better search experience and an even simpler listing process. If you have ideas for how we could improve things for you, or would like to help us make the magic happen, we’d love to hear from you!

In the meantime, here’s Lester from our Data Science squad talking about a bit of face modelling he’s been doing on the side.

This was originally posted on Default to Open

Author:

Simon Young, Head of Product Development, TradeMe

See Simon speak at the Chief Data & Analytics Officer New Zealand, 28-29 November, as he makes his keynote presentation - Making Life Better for Kiwis