Carsome

CARSOME
Prize - Internship for Fresh Graduate & Working Adult in Malaysia

Problem Statement: By looking at auction data, we needed your help to come up with a pricing algorithm for Carsome’s business that can accurately predict the ideal price range of the car, and them come up with a comprehensive data-driven sourcing strategy for identifying cars that should be purchased by Carsome.

Challenge I - Car Price Prediction

If you can estimate the price that a seller is willing to sell, or the price that a buyer is willing to buy, you can make your product or service more attractive by offering the right prices to your customers. Can we know at what price P, a car C can fetch in the market, at time T?

Hints:

What are the top car brands and models?
How does auction date or car mileage correlate with car prices?
Goal:

To predict the selling price of given cars in the inventory with the lowest error.

Challenge II - Data-Driven Sourcing Strategy

With a finite amount of sourcing budget, we need to squeeze the most value out of each dollar investment. Rather than using a blanket sourcing/procurement strategy, can we make a data driven decision to decide which set of cars to source and list?

Below are some possible suggestions to derive a better sourcing plan for Carsome’s retail business. .You are free to select one or more (or your own) suggestions to make the case. Feel free to come up with other suitable analysis / modeling based on the datasets provided.

Suggestion I - Dealer Search Analysis

Search, and you shall receive. With only 180 minutes of auction time and thousands of cars, dealers need efficient ways to scour through the auction to find great deals. Some of them use dropdown filters to explore possible options, but those with a clear idea of what they are looking for, may prefer a direct approach–keyword searches. Can we learn about a dealer’s preferences based on how and what they search?

Suggestion II - Car Segmentation

Segmentation is a crucial step in targeting the right market. It helps businesses make the right decision in delivering their winning strategy. Through many characteristics, interests, and behavior of our dealers, will we be able to tell what kind of Cars we should do business in?

Dataset: https://docs.google.com/spreadsheets/d/1xJTOc0hY5nj_LGj56VlcApf6VsV11iiEHsv8XJQ0V-E

Hint: Azure ML