Skip to content

Minimize Loss in Auto Auctions with IBM Predictive Analytics

Share This Post

In the next five years, revenue for the used car market is expected to trend higher. Because the industry relies strongly on consumers to drive revenue growth, fluctuations in disposable income dictate the direction of the industry. In the coming years, rising disposable income levels are anticipated to increase the likelihood of an individual purchasing a big-ticket item, such as a used automobile.

The Used Car industry is mature and relative to other automotive industries, has been fairly recession-resistant. Moreover, the industry has minimal barriers to entry, allowing individuals or non-employers to participate. The Used Car Dealers industry generally suffers from a poor public image as a result of the industry’s emphasis on high-margin sales techniques. Companies bucking this trend, such as CarMax, are rapidly gaining market share over traditional dealerships

Auto auction is one of the most important resources for dealers to purchase used cars. When purchasing a used car at an auto auction, one of the biggest challenges is that the vehicle might have serious issues that prevent the vehicle from being sold to customers. The auto community calls these unfortunate purchases “kicks” (bad buys). Bad-buy cars often result when there are tampered odometers, mechanical issues, or some other unforeseen problem. Bad-buy cars can be very costly to dealers.

Imagine you are in auto industry with used car selling as your core business, the ability to figure out which cars have a higher risk of being bad-buy can provide real value to your business. At this information age, with the emergence of big data, can predictive analysis do anything for the auto reselling business? The answer is yes, and we can help clients address these challenges using IBM SPSS Modeler – a predictive analytics solution.

trials-spss

Beginning by identifying critical attributes, such as auto age, make, odometer and auction price, the data was classified into either good buy segment or bad buy segment based on their characteristics. Then several algorithms were deployed in the model building process to gain a consolidated view of influencing factors.

With this solution, auto dealers would be able to learn which car model with specific attributes, for instance, a more than 5-year-old Ford Escape with greater than 110K miles, are more likely to be “kicks”. Now instead of taking actions after bad buys happened, the dealers have the capability to assess before buying cars from auctions and use other relevant information gained from this analysis to optimize decision making.

For more information on how you can use Predictive Analytics, connect with us today at crescointl.com

About The Author

Please enter you email to view this content.