Data Analysis on Used Car Sales
The skills I demoed here can be learned through taking Data Science with Machine Learning bootcamp with NYC Data Science Academy.
Shiny App
Introduction
All across the United States people get in a car and go somewhere on a daily basis. Whether the destination is work, the store or a friendβs house, the car is one of the primary means of transportation. In 2019 alone Americans drove over 3 trillion miles. Odds are good those miles were driven in a used car. In 2019, Americans bought 40 million used cars, more than twice as many as new cars. Based on data, the used car market is very important in the United States and making sure you have complete background knowledge of the market is very important for both buyers and sellers.
Looking around the internet I was able to find data on used car sales on eBay from 2017-2019. I dug into and analyzed what factors had the greatest impact on the price of a used car. Using R I was able to develop a Shiny App to help visualize the data.
Upon first acquiring the data the first step was to clean and filter the data to remove. anomalous data. Cars with very large mileage numbers such as over 1 billion or a year made in the 1800βs or after 2020 were immediately filtered out.The next category to beΒ filtered out wasΒ cars made before 1950. Cleaning the data involved adding a grouping of the decade a car was made, as well as grouping the mileage into groups of 25,000 miles up to 275,000.
DataΒ
Mileage and Year of the Car
The first piece of data I looked into was the impact of the mileage and year of the car. It became clear very quickly that both factors had a significant impact on the cost of a used car. Thereβs an inverse correlation between the mileage of a car and its sale price, and the price of a car increased for more recent models.
One interesting thing about this is while both trends generally held true there were outliers where the cost of a car would increase or hold steady even with high mileage numbers or being a much older car. This would make sense in cases where certain car models have value beyond just being a means of transportation, as many people collect cars so they may not be as concerned with mileage, and the older the car may be may have additional value to them. While I was not able to delve to much into this in my initial analysis it is something to look into more going forward.
Drive Type and Cylinders
Another area of interest is specific attributes of the car such as the drive type and number of cylinders in the car. When looking at these attributes there was not a correlation between the price of the car.
This makes sense as these features can be in cars that are new and old as well as having high and low mileage numbers.
Conclusion
While the data was valuable and I was able to extract many insights there is definitely room for additional exploration. Looking into the impact of collector cars as well as being able to see how specific car amenities such as GPS and music playing devices available could significantly aid the analysis.