CAPSTONE PROJECT
Nikhilan Velumani
December 29, 2017
Project Title : Analysis of online bidding in Ebay
NAME : Nikhilan Velumani
EMAIL : nikhilan.mbbs@gmail.com
COLLEGE / COMPANY: SRM University of Science and Techology
1. Introduction
An online auction is an auction which is held over the internet. Online auctions come in many different formats, but most popularly they are ascending English auctions, descending Dutch auctions, first-price sealed-bid, Vickrey auctions, or sometimes even a combination of multiple auctions, taking elements of one and forging them with another. The scope and reach of these auctions have been propelled by the Internet to a level beyond what the initial purveyors had anticipated.This is mainly because online auctions break down and remove the physical limitations of traditional auctions such as geography, presence, time, space, and a small target audience.This influx in reachability has also made it easier to commit unlawful actions within an auction.In 2002, online auctions were projected to account for 30% of all online e-commerce due to the rapid expansion of the popularity of the form of electronic commerce
2. Overview of the Study
This topic mainly foucus the study of e-auction which takes place on ebay. EBay is a publicly visible market which has attracted an interest from economists, who have used it to analyze aspects of buying and selling behavior, auction formats, etc., comparing them with previous theoretical and empirical findings.Computer information systems researchers have also shown interest in eBay. Michael Goul, Chairman of the Computer Information Systems department of the W. P. Carey School of Business at Arizona State University, published an academic case based on eBay’s big data management and use in which he discusses how eBay is a data-driven company that processes 50 petabytes of data a day.eBay uses a system that allows different departments in the company to check out data from their data mart into sandboxes for analysis. According to Goul, eBay has already experienced significant business successes through its data analytics. eBay employs 5,000 data analysts to enable data-driven decision making.
3. Study of online auction of Ebay
3.1 Overview
The objective of this study is to investigate behaviour and analaysis of bidding of several different persons for different products.We are going to analyze aspects of buying and selling behavior, auction formats, etc., comparing them with previous theoretical and empirical findings.
Hypothesis H1: During the auction the final selling prize of the object is always greater then the proxy bid
3.2 Data
For this study, we collected data from the http://www.modelingonlineauctions.com (http://www.modelingonlineauctions.com/code). What is the ‘Winner’s Curse’? The winner’s curse is a tendency for the winning bid in an auction to exceed the intrinsic value of the item purchased. Because of incomplete information, emotions or any other number of factors regarding the item being auctioned, bidders can have a difficult time determining the item’s intrinsic value. As a result, the largest overestimation of an item’s value ends up winning the auction.
3.3 Model
In order to test Hypothesis H1, we can propose the following model:
\[price= \alpha_0 + \alpha_1 bid + \alpha_2 openbid + \alpha_3 bidderrate \]
3.4 Read the data & Load the data file
bid.df <- read.csv(paste("CartierforWinnersCurse.csv", sep="new"))details of each field
auctionid - unique identifier of an auction
bid - the proxy bid placed by a bidder
bidtime - the time (in days) that the bid was placed, from the start of the auction
bidder - eBay username of the bidder
bidderrate - eBay feedback rating of the bidder
openbid - the opening bid set by the seller
price - the closing price that the item sold for (equivalent to the second highest bid + an increment)
View the data
View(bid.df)to find the length and breadth of your dataset.
library(psych)
attach(bid.df)
describe(bid.df)[,c(1:5)]## vars n mean sd median
## auctionid 1 1348 1.644515e+09 3597127.86 1.644198e+09
## bid 2 1348 5.985700e+02 659.81 3.530000e+02
## bidtime 3 1348 4.010000e+00 2.50 4.170000e+00
## bidder* 4 1348 2.576500e+02 148.11 2.675000e+02
## bidderrate 5 1348 3.386000e+01 87.51 6.000000e+00
## openbid 6 1348 1.486800e+02 373.16 5.000000e+00
## price 7 1348 9.613100e+02 812.41 6.232600e+02summary of data
summary(bid.df)[,c(1:5)]## auctionid bid bidtime
## Min. :1.639e+09 Min. : 1.0 Min. :0.007535
## 1st Qu.:1.642e+09 1st Qu.: 151.7 1st Qu.:1.505715
## Median :1.644e+09 Median : 353.0 Median :4.170885
## Mean :1.645e+09 Mean : 598.6 Mean :4.005524
## 3rd Qu.:1.648e+09 3rd Qu.: 821.5 3rd Qu.:6.725284
## Max. :1.651e+09 Max. :5400.0 Max. :6.999965
##
## bidder bidderrate
## lass1004 : 22 Min. : -4.00
## pascal1666 : 19 1st Qu.: 1.00
## freembd : 17 Median : 6.00
## happyrova : 17 Mean : 33.86
## restdynamics: 17 3rd Qu.: 31.00
## adammurry : 16 Max. :1303.00
## (Other) :12403.4 Results
We can see through our further analysis , we can say that the final selling price of the product is always greater then the proxy bid.
4 Conclusion
This project was done to analyse the winners curse effect among the bidders. Which is our sole hypothesis which is proved. This proect was done to investigate behaviour and analaysis of bidding of several different persons for different products.We are going to analyze aspects of buying and selling behavior, auction formats, etc., comparing them with previous theoretical and empirical findings.
5. References
wikipedia.org , the details of ebay ,Available from: https://en.wikipedia.org/wiki/EBay
Details about the winners Curse ,
Available from:https://www.investopedia.com/terms/w/winnerscurse.asp
Data take from modelingonlineauctions website , Available from:http://www.modelingonlineauctions.com
Analysis of Total projects
summary of Auction id
describe(auctionid)## vars n mean sd median trimmed mad min
## X1 1 1348 1644515087 3597128 1644197869 1644463742 4642817 1638843936
## max range skew kurtosis se
## X1 1650986455 12142519 0.18 -1.12 97974.02summary of the proxy bid placed by a bidder
describe(bid)## vars n mean sd median trimmed mad min max range skew
## X1 1 1348 598.57 659.81 353 482.26 375.1 1 5400 5399 2.32
## kurtosis se
## X1 8.69 17.97summary of the time (in days) that the bid was placed, from the start of the auction
describe(bidtime)## vars n mean sd median trimmed mad min max range skew kurtosis
## X1 1 1348 4.01 2.5 4.17 4.09 3.82 0.01 7 6.99 -0.14 -1.57
## se
## X1 0.07summary of bidderrate - eBay feedback rating of the bidder
describe(bidderrate)## vars n mean sd median trimmed mad min max range skew kurtosis
## X1 1 1348 33.86 87.51 6 15.5 8.9 -4 1303 1307 7.59 82.65
## se
## X1 2.38summary of openbid - the opening bid set by the seller
describe(openbid)## vars n mean sd median trimmed mad min max range skew
## X1 1 1348 148.68 373.16 5 73.91 7.4 0.01 5000 4999.99 7.56
## kurtosis se
## X1 85.25 10.16summary of price - the closing price that the item sold for
describe(price)## vars n mean sd median trimmed mad min max range skew
## X1 1 1348 961.31 812.41 623.26 840.09 545.98 103.5 5400 5296.5 1.53
## kurtosis se
## X1 3.01 22.13histogram representation of Price field
library(lattice)
histogram(~price, data = bid.df,
main = "Distribution of Price Difference", xlab="Difference in Price", col='gray' )
Scatterplot of Price vs Openbid
library(car)##
## Attaching package: 'car'## The following object is masked from 'package:psych':
##
## logitlibrary(psych)
scatterplot(price ~ openbid ,data=bid.df, main="Scatterplot of Price vs Openbid ")
Scatterplot of proxy bid or first bid vs Openbid
scatterplot(bid ~ openbid ,data=bid.df, main="Scatterplot of Price vs Openbid ")
Scatterplot of all the required elements
library(car)
scatterplotMatrix(~openbid+price+bid+bidderrate+bidtime, data=bid.df,
main="Premium Economy vs. Economy Airfares")
CORRELATION MATRIX
library(Hmisc)## Warning: package 'Hmisc' was built under R version 3.4.3## Loading required package: survival## Loading required package: Formula## Loading required package: ggplot2##
## Attaching package: 'ggplot2'## The following objects are masked from 'package:psych':
##
## %+%, alpha##
## Attaching package: 'Hmisc'## The following object is masked from 'package:psych':
##
## describe## The following objects are masked from 'package:base':
##
## format.pval, round.POSIXt, trunc.POSIXt, unitscolbid <- c("price","openbid","bid","bidderrate")
corMatrix <- rcorr(as.matrix(bid.df[,colbid]))
corMatrix## price openbid bid bidderrate
## price 1.00 0.42 0.82 -0.07
## openbid 0.42 1.00 0.56 0.02
## bid 0.82 0.56 1.00 -0.04
## bidderrate -0.07 0.02 -0.04 1.00
##
## n= 1348
##
##
## P
## price openbid bid bidderrate
## price 0.0000 0.0000 0.0087
## openbid 0.0000 0.0000 0.4164
## bid 0.0000 0.0000 0.1759
## bidderrate 0.0087 0.4164 0.1759CORRGRAM of biddings
library(Hmisc)
library(car)
library(corrgram)## Warning: package 'corrgram' was built under R version 3.4.3colbid <- c("price","openbid","bid","bidderrate")
corrgram(bid.df[,colbid], order=TRUE,
main="Difference in biddings",
lower.panel=panel.pts, upper.panel=panel.pie,
diag.panel=panel.minmax, text.panel=panel.txt)
to check that the the Price of the object sold is always greateer then theproxy bid
t.test(price, bid)##
## Welch Two Sample t-test
##
## data: price and bid
## t = 12.725, df = 2585.3, p-value < 2.2e-16
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## 306.8423 418.6353
## sample estimates:
## mean of x mean of y
## 961.3078 598.5690