Autoworth is a website that helps people analyze used vehicle listings. The user enters the vehicle make, model, year, asking price, mileage, and an optional VIN. The website then gives back information such as the predicted price, average market price, reliability estimate, and a short recommendation. The project used a vehicle dataset with information like manufacturer, model, price, odometer reading, year, vehicle type, generation, and reliability values. The main purpose was to take raw vehicle data and make it easier for a user to understand whether a used car listing looks reasonable.
One Industrial Engineering idea connected to our project was reliability analysis. We used Mean Time To Failure (MTTF) and Weibull reliability values to estimate how much life a vehicle may have left based on its mileage. This was useful because a car can have a fair price but still be risky if it is close to a major failure point. Some of the skills we practiced included using external libraries like SMILE and Vaadin, connecting to an external API, and using GitHub to work on the project as a team.
Project Development Process
Our original goal was to make a program that could help users decide if a used vehicle was a good deal. At first, the project was mostly about comparing a user’s price to vehicle prices in the dataset. As we worked on it, we added more features, including reliability estimates and VIN decoding. The project changed from a basic Java analyzer into an accurate reliability estimator and predictor. The backend handled the vehicle calculations, while the website gave users a cleaner way to enter information and read the results. One roadblock was working with vehicle data that was not always perfectly formatted. Model names, trims, and vehicle details could be inconsistent, so the program needed logic to clean and match user inputs with the dataset. Especially the different trim models that did not match across data sets. Another challenge was connecting different parts of the project together, such as the price model, reliability calculations, VIN decoder, and user interface.
Key Features
One important feature was the price prediction model. The project used a SMILE random forest model to estimate a vehicle’s market value based on information like make, model, year, mileage, and vehicle type. Another important feature was the reliability estimate. The program used Weibull reliability values to calculate an estimated Mean Time To Failure. This helped the website explain whether a vehicle may still have useful life left or whether the mileage could be a concern. I worked almost entirely on the front-end of the project ensuring that all our features were visible and usable on the vaadin page. There were a lot of vaadin components to build though most of them, like the combo boxes, were quite repetitive. The most challenging parts of this project was getting all of our vertical layout components to fit to the screen properly and get our header to quit taking up half the screen.
This example shows the app analyzing a 2019 Volkswagen Golf. The VIN decoder adds extra details such as trim, body style, fuel type, and transmission, while the rest of the app displays price, reliability, survival chance, and decision guidance. The website also gives the user a final recommendation based on all of the metrics provided. Instead of only showing numbers, it explains if the vehicle looks like a good deal, if it is above market value, or if the reliability estimate should make the buyer more cautious.
Reflection
I think that this project gave me a lot of insight into front-end development and ui design, we could make the most interesting and useful tool on the planet but if it is not easily useable then we might as well have nothing at all. This project showed me how incredibly diverse ui design can be, I also got to learn some more technical things like how to use a data service and why they are important to projects like ours. I believe these hard skills in the development and implementation of the concepts we have been learning will help out a great amount in future projects and eventually our careers.