1718t1is428T11

PROPOSAL

POSTER

APPLICATION

RESEARCH PAPER

Introduction

According to Forbes, the automobile industry has grown by a massive 68% since hitting a trough during the 2009 global financial crises according to a report published by car auction company Manheim earlier this year.

Q3 2016 closed with 9.8M vehicles sold in the used car market -an increase of 3.3% over the previous year. Also the average retail used vehicle sold for $19.232 in Q3 2016, an increase of 4.3% over last year. Changes in car buying behavior are beginning to alter the landscape of franchised used vehicles.

So both franchised used car firms and other giant online marketplaces like E-Bay are leveraging the growth rate of used car industy. As a result, we tried to understand this market and its dynamics with the help of ‘Used Car Database’ from E-Bay.

Problem and Motivation

With the change in consumer car buying behaviour and a rising market for used cars, our aim is to understand this growing used car market. When consumers look at used cars, price is the most important factor that influences opinions, but there are also few other facts that affect their purchasing decision. So we will try to find out which variables affect the price most and how they are correlated.

Objective

In this project, we are interested to create a visualisation application that helps users perform the following:

1. Visualise resale car prices against other factors such as:

• Type of cars
• Car Brands
• Car age
• Kilometers traveled
• Time period
• Location/ state

2. Identify relationships and correlations across different factors affecting resale prices

3. Uncover the top 10 most common brands for car resale

• Difference in prices & quantity sold across different brands

Data

Data used is obtained from Kaggle website and it is about used cars in Ebay Kleinanzeigen. This dataset contains 371539 records and consists of following columns.

Attribute	Description
dateCrawled	when this ad was first crawled, all field-values are taken from this date
name	"name" of the car
seller	private or dealer
fuelType	"name" of the car
price	the price on the ad to sell the car
vehicleType	vehicle type
yearOfRegistration	at which year the car was first registered
gearbox	gearbox
brand	brand of the car
notRepairedDamage	if the car has a damage which is not repaired yet
dateCreated	the date for which the ad at ebay was created
nrOfPictures	number of pictures in the ad
lastSeenOnline	when the crawler saw this ad last online
powerPS	power of the car in PS
km	how many kilometers the car has driven
model	car model
monthOfRegistration	at which month the car was first registered
postalCode	postal code
offerType	/
abtest	/

Research Visualisation

Visualizations	Explaination
	Bubble Chart This figure allows us to visualize vehicle types and sales at the same time. This chart is very interactive as well. Readers can group/color the data points by “major brand”, “origin”, “truck/car” and “gainers/losers”. It provides comprehensive insights about vehicle sales with a straightforward visualization. https://www.bloomberg.com/graphics/2015-auto-sales/
	Scatter Plot Scatter plot can help readers to understand the relationship between two variables. This visualization shows how age and odometer are correlated with car price. In our case, we would like to investigate which factors contribute to the price of the used cars. http://csidsocialmedia.github.io/2014/05/02/Predict-second-hand-car-price-using-artificial-neural-network.html
	Line Graph This line graph is useful for displaying the price changes of the used cars over time. Although this graph is not interactive, it helps the readers to detect patterns or trends. Line graph is also good at comparison. In our case, we could use the line graph to compare the price changes of different car models to discover useful insights. http://www.zerohedge.com/news/2017-05-21/perfect-storm-hits-used-car-values-foundation-auto-industry-faltering

Tools

 -Excel

 -D3

 -Javascript

 -Github

 -Tableau

Technical Challenges

Technical Challenges	Action Plan
Data Preparation	Work on data cleaning and transforming.
Unfamiliarity in Programming Language like Javascript & Libraries like D3	Initial hands-on experience during D3.js workshop. Independent learning on Javascript & D3.js. Peer learning and sharing of skills.
Unfamiliarity in Implementing Interactive Visualisation App	Self-learning and view online tutorials.

Roles & Milestones

• Project Roles

Dong Ruiyan: Visualisation Analyst
Zhang Qian: Visualisation Designer
Jeremy LEE Ting Kok: Project Manager

• Project Timeline

References

• Bloomberg - Scientific Proof that Americans are Completely Addicted to Trucks: https://www.bloomberg.com/graphics/2015-auto-sales/
• Predict second hand car price using artificial neural network: http://csidsocialmedia.github.io/2014/05/02/Predict-second-hand-car-price-using-artificial-neural-network.html
• The Perfect Storm Hits Used-Car Values: The Foundation Of The Auto Industry Is Faltering: http://www.zerohedge.com/news/2017-05-21/perfect-storm-hits-used-car-values-foundation-auto-industry-faltering
• Kaggle - Used car database: https://www.kaggle.com/orgesleka/used-cars-database
• Kaggle - Data Crunchers: https://www.kaggle.com/timucinanuslu/data-crunchers
• D3.js: https://d3js.org/
• Chart.js: http://www.chartjs.org/