Estimating the optimal price of an item using AI algorithms

The dedicated auction system is a simple, functional and modern solution that allows selling and buying goods through auctions or direct purchases within a closed group of users (e.g. employees or customers). It is a modern e-commerce solution that allows you to create internal auctions for a specific group of people.

Industry

e-commerce trade

Country

Poland

Technologies

Django Python Scikit-Learn

Scope

integrations mobile application web application

Challenge

Determining the right resale price for used products, especially electronics, is often complicated. The products vary in many parameters and have different specifications. In the case of items that have been withdrawn from official sales channels, the challenge becomes even more difficult.

Solution

We have developed a mechanism for estimating product prices using laptops as an example. Each laptop is described by several key parameters, and additional information (such as brand or model) often appears in the listing name. We used machine learning methods to analyse prices, testing different mathematical models depending on the data available and the specifics of the market.

Linear models were able to identify patterns quickly, even with small data sets, and presented the results in a clear way.
Neural networks required larger amounts of data, but provided predictions that were much closer to what was actually on the market.
KNN (k-Nearest Neighbours) algorithms allowed you to see offers that were most similar to the product you entered, helping retailers to manually adjust the price.

Benefits for users:

Sellers get an estimate of the most appropriate price – after entering the parameters of their product, they can see the suggested price based on the most recent market offers.
Buyers can instantly check the attractiveness of a particular listing – each listing is benchmarked against similar listings, making it easier to assess whether the price is fair, a bargain or overpriced.
Search for the best deals – buyers can see a list of the most best deals according to their search criteria.
List of products with similar specifications – buyers can quickly find and compare listings with similar specifications.
Price Driver Analysis – buyers can easily identify which parameters have the greatest impact on price and adjust them to suit their needs if necessary.

Business impact

The introduction of AI into price estimation gives users direct access to information that was previously difficult to obtain. This in turn saves time for both sellers and buyers. Sellers can quickly determine the right price for their listing without having to manually compare products on the market. Buyers get a tool to instantly compare offers, simplifying the buying process. This makes the user experience on the platform much friendlier, leading to increased frequency and satisfaction with the site.