Peeking Beneath the Hood of Uber Le Chen, Alan Mislove, Christo - - PowerPoint PPT Presentation

▶
peeking beneath the hood of uber
SMART_READER_LITE
LIVE PREVIEW

Peeking Beneath the Hood of Uber Le Chen, Alan Mislove, Christo - - PowerPoint PPT Presentation

Sep 19, 2023 290 likes •2.15k views

Peeking Beneath the Hood of Uber Le Chen, Alan Mislove, Christo Wilson Northeastern University What is Uber? 2 What is Uber? 2 What is Uber? Type pickup location 2 What is Uber? Type pickup location 2 What is Uber? Type pickup location

slide-1
SLIDE 1

Peeking Beneath the Hood of Uber

Le Chen, Alan Mislove, Christo Wilson Northeastern University

slide-2
SLIDE 2

What is Uber?

2

slide-3
SLIDE 3

What is Uber?

2

slide-4
SLIDE 4

What is Uber?

Type pickup location

2

slide-5
SLIDE 5

What is Uber?

Type pickup location

2

slide-6
SLIDE 6

What is Uber?

Type pickup location Choose type of car

2

slide-7
SLIDE 7

What is Uber?

Type pickup location Choose type of car

  • UberX: basic sedans

2

slide-8
SLIDE 8

What is Uber?

Type pickup location Choose type of car

  • UberX: basic sedans
  • UberBlack: luxury cars

2

slide-9
SLIDE 9

What is Uber?

3

slide-10
SLIDE 10

What is Uber?

3

slide-11
SLIDE 11

What is Uber?

Request!

3

slide-12
SLIDE 12

Simple and smooth! But …

4

slide-13
SLIDE 13

Simple and smooth! But …

4

slide-14
SLIDE 14

Blackbox algorithm

5

slide-15
SLIDE 15

Blackbox algorithm

5

  • Transparent marketplaces
slide-16
SLIDE 16

Blackbox algorithm

5

  • Transparent marketplaces
slide-17
SLIDE 17

Blackbox algorithm

Prices are set by the suppliers

5

  • Transparent marketplaces
slide-18
SLIDE 18

Blackbox algorithm

Prices are set by the suppliers

  • Uber is not a transparent marketplace

5

  • Transparent marketplaces
slide-19
SLIDE 19

Blackbox algorithm

Prices are set by the suppliers

  • Uber is not a transparent marketplace

Prices are dynamically calculated by an algorithm based on supply, demand, etc.

5

  • Transparent marketplaces
slide-20
SLIDE 20

Blackbox algorithm

6

slide-21
SLIDE 21

Blackbox algorithm

6

slide-22
SLIDE 22

Blackbox algorithm

6

slide-23
SLIDE 23

Blackbox algorithm

6

slide-24
SLIDE 24

Blackbox algorithm

6

slide-25
SLIDE 25

Blackbox algorithm

Popular, but not transparent

6

slide-26
SLIDE 26

Goal of this study

7

slide-27
SLIDE 27

Goal of this study

  • How does the surge pricing algorithm work?

7

slide-28
SLIDE 28

Goal of this study

  • How does the surge pricing algorithm work?
  • Verify if it works as Uber claimed

7

slide-29
SLIDE 29

Goal of this study

  • How does the surge pricing algorithm work?
  • Verify if it works as Uber claimed
  • Responsiveness of the algorithm

7

slide-30
SLIDE 30

Goal of this study

  • How does the surge pricing algorithm work?
  • Verify if it works as Uber claimed
  • Responsiveness of the algorithm
  • Can the surge values be predicted?

7

slide-31
SLIDE 31

Goal of this study

  • How does the surge pricing algorithm work?
  • Verify if it works as Uber claimed
  • Responsiveness of the algorithm
  • Can the surge values be predicted?
  • Impact on passengers and drivers

7

slide-32
SLIDE 32

Outline of Uber study

  • Motivation
  • Data collection
  • Surge pricing
  • Summary
slide-33
SLIDE 33

Data collection

9

slide-34
SLIDE 34

Data collection

  • Uber official patent: supply, demand, etc.

9

slide-35
SLIDE 35

Data collection

  • Uber official patent: supply, demand, etc.
  • Option 1: Uber API

9

slide-36
SLIDE 36

Data collection

  • Uber official patent: supply, demand, etc.
  • Option 1: Uber API
  • Pros: easy to use; has surge values and EWT

9

slide-37
SLIDE 37

Data collection

  • Uber official patent: supply, demand, etc.
  • Option 1: Uber API
  • Pros: easy to use; has surge values and EWT
  • Cons: no car/demand/supply information

9

slide-38
SLIDE 38

Data collection

  • Uber official patent: supply, demand, etc.
  • Option 1: Uber API
  • Pros: easy to use; has surge values and EWT
  • Cons: no car/demand/supply information
  • Option 2: Uber client app

9

slide-39
SLIDE 39

Uber client app

10

slide-40
SLIDE 40

Uber client app

10

slide-41
SLIDE 41

Uber client app

  • Pings the server every 5 seconds

10

slide-42
SLIDE 42

Uber client app

  • Pings the server every 5 seconds
  • 8 nearest cars

10

slide-43
SLIDE 43

Uber client app

  • Pings the server every 5 seconds
  • 8 nearest cars
  • Estimated Wait Time (EWT)

10

slide-44
SLIDE 44

Uber client app

  • Pings the server every 5 seconds
  • 8 nearest cars
  • Estimated Wait Time (EWT)
  • Surge multiplier

10

slide-45
SLIDE 45

Uber client app

  • Pings the server every 5 seconds
  • 8 nearest cars
  • Estimated Wait Time (EWT)
  • Surge multiplier
  • More information (supply/demand)

10

slide-46
SLIDE 46

Uber client app limitations

11

slide-47
SLIDE 47

Uber client app limitations

  • Supply and demand are estimations

11

slide-48
SLIDE 48

Uber client app limitations

  • Supply and demand are estimations
  • Supply = number of cars on the road

11

slide-49
SLIDE 49

Uber client app limitations

  • Supply and demand are estimations
  • Supply = number of cars on the road
  • Fulfilled demand = number of cars

go offline

11

slide-50
SLIDE 50

Uber client app limitations

  • Supply and demand are estimations
  • Supply = number of cars on the road
  • Fulfilled demand = number of cars

go offline

11

slide-51
SLIDE 51

Uber client app limitations

  • Supply and demand are estimations
  • Supply = number of cars on the road
  • Fulfilled demand = number of cars

go offline

  • Got booked

11

slide-52
SLIDE 52

Uber client app limitations

  • Supply and demand are estimations
  • Supply = number of cars on the road
  • Fulfilled demand = number of cars

go offline

  • Got booked
  • Logged out

11

slide-53
SLIDE 53

Uber client app limitations

  • Supply and demand are estimations
  • Supply = number of cars on the road
  • Fulfilled demand = number of cars

go offline

  • Got booked
  • Logged out
  • Driving out of boundary

11

slide-54
SLIDE 54

Uber client app limitations

  • Supply and demand are estimations
  • Supply = number of cars on the road
  • Fulfilled demand = number of cars

go offline

  • Got booked
  • Logged out
  • Driving out of boundary

Upper bound of fulfilled demand!

11

slide-55
SLIDE 55

Uber client app limitations

12

slide-56
SLIDE 56

Uber client app limitations

  • Limited visibility

12

slide-57
SLIDE 57

Uber client app limitations

  • Limited visibility
  • Only 8 nearest cars

12

slide-58
SLIDE 58

Uber client app limitations

  • Limited visibility
  • Only 8 nearest cars
  • Limited number of measuring points

12

slide-59
SLIDE 59

Uber client app limitations

  • Limited visibility
  • Only 8 nearest cars
  • Limited number of measuring points
  • How far away between points?

12

slide-60
SLIDE 60

Limited visibility

13

slide-61
SLIDE 61

Limited visibility

13

slide-62
SLIDE 62

Limited visibility

13

slide-63
SLIDE 63

Limited visibility

13

slide-64
SLIDE 64

Limited visibility

13

slide-65
SLIDE 65

Limited visibility

13

Radius r

slide-66
SLIDE 66

Limited visibility

14

slide-67
SLIDE 67

Limited visibility

14

slide-68
SLIDE 68

Limited visibility

14

slide-69
SLIDE 69

Limited visibility

14

slide-70
SLIDE 70

Limited visibility

Radius r

14

slide-71
SLIDE 71

Radius measurement

15

slide-72
SLIDE 72

Radius measurement

  • Measuring radius (details in paper)

15

slide-73
SLIDE 73

Radius measurement

  • Measuring radius (details in paper)
  • Midtown Manhattan: 247m

15

slide-74
SLIDE 74

Radius measurement

  • Measuring radius (details in paper)
  • Midtown Manhattan: 247m
  • Downtown San Francisco: 387m

15

slide-75
SLIDE 75

Radius measurement

  • Measuring radius (details in paper)
  • Midtown Manhattan: 247m
  • Downtown San Francisco: 387m
  • Use the NYC taxi data to validate that 97% of supply and

95% of the demand are covered

15

slide-76
SLIDE 76

Radius measurement

  • Measuring radius (details in paper)
  • Midtown Manhattan: 247m
  • Downtown San Francisco: 387m
  • Use the NYC taxi data to validate that 97% of supply and

95% of the demand are covered

  • NYC and SF are the targets for collecting data

15

slide-77
SLIDE 77

Radius measurement

  • Measuring radius (details in paper)
  • Midtown Manhattan: 247m
  • Downtown San Francisco: 387m
  • Use the NYC taxi data to validate that 97% of supply and

95% of the demand are covered

  • NYC and SF are the targets for collecting data
  • 2nd and 3rd largest Uber market

15

slide-78
SLIDE 78

Radius measurement

  • Measuring radius (details in paper)
  • Midtown Manhattan: 247m
  • Downtown San Francisco: 387m
  • Use the NYC taxi data to validate that 97% of supply and

95% of the demand are covered

  • NYC and SF are the targets for collecting data
  • 2nd and 3rd largest Uber market
  • Different access to public transportation

15

slide-79
SLIDE 79

The measurement grid

16

slide-80
SLIDE 80

The measurement grid

16

slide-81
SLIDE 81

The measurement grid

16

slide-82
SLIDE 82

Ethics

17

slide-83
SLIDE 83

Ethics

  • We did not collect personal information of

passengers and Uber drivers

17

slide-84
SLIDE 84

Ethics

  • We did not collect personal information of

passengers and Uber drivers

  • We never booked Uber rides

17

slide-85
SLIDE 85

Ethics

  • We did not collect personal information of

passengers and Uber drivers

  • We never booked Uber rides
  • We did not induce surges

17

slide-86
SLIDE 86

Outline of Uber study

  • Motivation
  • Data collection
  • Surge pricing
  • Summary
slide-87
SLIDE 87

Research questions

19

slide-88
SLIDE 88

Research questions

  • How much and how often does it surge on Uber?

19

slide-89
SLIDE 89

Research questions

  • How much and how often does it surge on Uber?
  • How long do surge lasts?

19

slide-90
SLIDE 90

Research questions

  • How much and how often does it surge on Uber?
  • How long do surge lasts?
  • How do surge prices vary by location?

19

slide-91
SLIDE 91

Research questions

  • How much and how often does it surge on Uber?
  • How long do surge lasts?
  • How do surge prices vary by location?
  • What features does Uber use to calculate surge

multipliers?

19

slide-92
SLIDE 92

Research questions

  • How much and how often does it surge on Uber?
  • How long do surge lasts?
  • How do surge prices vary by location?
  • What features does Uber use to calculate surge

multipliers?

  • Can we predict surges? If not, can we avoid surges?

19

slide-93
SLIDE 93

Research questions

  • How much and how often does it surge on Uber?
  • How long do surge lasts?
  • How do surge prices vary by location?
  • What features does Uber use to calculate surge

multipliers?

  • Can we predict surges? If not, can we avoid surges?
  • What is the impact of surge pricing?

19

slide-94
SLIDE 94

How much and how often does it surge on Uber?

20

slide-95
SLIDE 95

How much and how often does it surge on Uber?

CDF 20 40 60 80 100 Surge multiplier 1 2 3 4 5

Manhattan San Francisco

20

slide-96
SLIDE 96

How much and how often does it surge on Uber?

CDF 20 40 60 80 100 Surge multiplier 1 2 3 4 5

Manhattan San Francisco

20

slide-97
SLIDE 97

How much and how often does it surge on Uber?

CDF 20 40 60 80 100 Surge multiplier 1 2 3 4 5

Manhattan San Francisco

20

  • 14% of time surging in Manhattan
slide-98
SLIDE 98

How much and how often does it surge on Uber?

CDF 20 40 60 80 100 Surge multiplier 1 2 3 4 5

Manhattan San Francisco

20

  • 14% of time surging in Manhattan
slide-99
SLIDE 99

How much and how often does it surge on Uber?

CDF 20 40 60 80 100 Surge multiplier 1 2 3 4 5

Manhattan San Francisco

20

  • 14% of time surging in Manhattan
  • 57% of time surging in San Francisco
slide-100
SLIDE 100

How much and how often does it surge on Uber?

CDF 20 40 60 80 100 Surge multiplier 1 2 3 4 5

Manhattan San Francisco

20

  • 14% of time surging in Manhattan
  • 57% of time surging in San Francisco
  • SF has higher surge values
slide-101
SLIDE 101

How long do surges last?

21

slide-102
SLIDE 102

How long do surges last?

CDF 20 40 60 80 100 Surge duration 5 mins 10 mins 1 hour 1 min

21

slide-103
SLIDE 103

How long do surges last?

CDF 20 40 60 80 100 Surge duration 5 mins 10 mins 1 hour 1 min

21

slide-104
SLIDE 104

How long do surges last?

CDF 20 40 60 80 100 Surge duration 5 mins 10 mins 1 hour 1 min

  • Noisiness: 70% of surges last less than 10 mins

21

slide-105
SLIDE 105

How long do surges last?

CDF 20 40 60 80 100 Surge duration 5 mins 10 mins 1 hour 1 min

  • Noisiness: 70% of surges last less than 10 mins
  • Staircase CDF: multiple times of 5 minutes

21

slide-106
SLIDE 106

How long do surges last?

CDF 20 40 60 80 100 Surge duration 5 mins 10 mins 1 hour 1 min

  • Noisiness: 70% of surges last less than 10 mins
  • Staircase CDF: multiple times of 5 minutes
  • Uber updates surge values every 5 minutes

21

slide-107
SLIDE 107

How do surge prices vary by location?

slide-108
SLIDE 108

How do surge prices vary by location?

slide-109
SLIDE 109

How do surge prices vary by location?

slide-110
SLIDE 110

How do surge prices vary by location?

slide-111
SLIDE 111

How do surge prices vary by location?

slide-112
SLIDE 112

How do surge prices vary by location?

slide-113
SLIDE 113

How do surge prices vary by location?

slide-114
SLIDE 114

Surge areas

26

slide-115
SLIDE 115

Surge areas

  • Chicago
  • Los Angeles
  • Miami
  • Seattle
  • NYC, including all 5 boroughs
  • Boston
  • DC
  • London
  • Paris

26

slide-116
SLIDE 116

What features does Uber use to calculate surge multipliers?

27

slide-117
SLIDE 117

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing

27

slide-118
SLIDE 118

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.

27

slide-119
SLIDE 119

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.
  • Cross-correlation

27

slide-120
SLIDE 120

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.
  • Cross-correlation

27

Time

Surge Var

slide-121
SLIDE 121

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.
  • Cross-correlation

27

Time

Surge Var

slide-122
SLIDE 122

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.
  • Cross-correlation

27

Time

Surge Var

slide-123
SLIDE 123

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.
  • Cross-correlation

27

Time

Surge Var

5 minute window

slide-124
SLIDE 124

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.
  • Cross-correlation

27

Time

Surge Var

5 minute window

Time Difference = 0

slide-125
SLIDE 125

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.
  • Cross-correlation

27

Time

Surge Var

Time Difference = 0

slide-126
SLIDE 126

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.
  • Cross-correlation

27

Time

Surge Var

Time Difference = 0

slide-127
SLIDE 127

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.
  • Cross-correlation

27

Time

Surge Var

Time Difference = -5

slide-128
SLIDE 128

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.
  • Cross-correlation

27

Time

Surge Var

Time Difference = -5

slide-129
SLIDE 129

What features does Uber use to calculate surge multipliers?

  • Find variables that correlate with surge pricing
  • Supply, demand, EWT, etc.
  • Cross-correlation

27

Time

Surge Var

Time Difference = +5

slide-130
SLIDE 130

(Supply - demand) vs surge multipliers

28

slide-131
SLIDE 131

(Supply - demand) vs surge multipliers

  • 0.6
  • 0.4
  • 0.2

0.2 0.4

  • 60
  • 40
  • 20

20 40 60 0.2 0.4 0.6 0.8 1 Correlation Coefficient Probability Time Difference in Minutes NYC Correlation NYC p-value SFCorrelation NYC p-value

28

slide-132
SLIDE 132

(Supply - demand) vs surge multipliers

  • 0.6
  • 0.4
  • 0.2

0.2 0.4

  • 60
  • 40
  • 20

20 40 60 0.2 0.4 0.6 0.8 1 Correlation Coefficient Probability Time Difference in Minutes NYC Correlation NYC p-value SFCorrelation NYC p-value

28

slide-133
SLIDE 133

EWT vs surge multipliers

29

slide-134
SLIDE 134

EWT vs surge multipliers

  • 0.4
  • 0.2

0.2 0.4 0.6

  • 60
  • 40
  • 20

20 40 60 0.2 0.4 0.6 0.8 1 Correlation Coefficient Probability Time Difference in Minutes NYC Correlation NYC p-value SFCorrelation NYC p-value

29

slide-135
SLIDE 135

EWT vs surge multipliers

  • 0.4
  • 0.2

0.2 0.4 0.6

  • 60
  • 40
  • 20

20 40 60 0.2 0.4 0.6 0.8 1 Correlation Coefficient Probability Time Difference in Minutes NYC Correlation NYC p-value SFCorrelation NYC p-value

29

slide-136
SLIDE 136

Features correlated with surge multipliers

30

slide-137
SLIDE 137

Features correlated with surge multipliers

  • 0.4
  • 0.2

0.2 0.4 0.6

  • 60
  • 40
  • 20

20 40 60 0.2 0.4 0.6 0.8 1 Correlation Coefficient Probability Time Difference in Minutes NYC Correlation NYC p-value SFCorrelation NYC p-value

  • 0.6
  • 0.4
  • 0.2

0.2 0.4

  • 60
  • 40
  • 20

20 40 60 0.2 0.4 0.6 0.8 1 Correlation Coefficient Probability Time Difference in Minutes NYC Correlation NYC p-value SFCorrelation NYC p-value

(Supply - demand) vs surge EWT vs surge

30

slide-138
SLIDE 138

Features correlated with surge multipliers

  • 0.4
  • 0.2

0.2 0.4 0.6

  • 60
  • 40
  • 20

20 40 60 0.2 0.4 0.6 0.8 1 Correlation Coefficient Probability Time Difference in Minutes NYC Correlation NYC p-value SFCorrelation NYC p-value

  • 0.6
  • 0.4
  • 0.2

0.2 0.4

  • 60
  • 40
  • 20

20 40 60 0.2 0.4 0.6 0.8 1 Correlation Coefficient Probability Time Difference in Minutes NYC Correlation NYC p-value SFCorrelation NYC p-value

(Supply - demand) vs surge EWT vs surge

  • Moderate correlation when time difference is 0

30

slide-139
SLIDE 139

Features correlated with surge multipliers

  • 0.4
  • 0.2

0.2 0.4 0.6

  • 60
  • 40
  • 20

20 40 60 0.2 0.4 0.6 0.8 1 Correlation Coefficient Probability Time Difference in Minutes NYC Correlation NYC p-value SFCorrelation NYC p-value

  • 0.6
  • 0.4
  • 0.2

0.2 0.4

  • 60
  • 40
  • 20

20 40 60 0.2 0.4 0.6 0.8 1 Correlation Coefficient Probability Time Difference in Minutes NYC Correlation NYC p-value SFCorrelation NYC p-value

(Supply - demand) vs surge EWT vs surge

  • Moderate correlation when time difference is 0
  • Zero correlation in other windows: responsive but

noisy

30

slide-140
SLIDE 140

Can we predict surge multipliers?

31

slide-141
SLIDE 141

Can we predict surge multipliers?

  • Useful variables

31

slide-142
SLIDE 142

Can we predict surge multipliers?

  • Useful variables
  • Supply/demand difference, EWT, surge history

31

slide-143
SLIDE 143

Can we predict surge multipliers?

  • Useful variables
  • Supply/demand difference, EWT, surge history
  • Model

31

slide-144
SLIDE 144

Can we predict surge multipliers?

  • Useful variables
  • Supply/demand difference, EWT, surge history
  • Model
  • Linear regression

31

slide-145
SLIDE 145

Can we predict surge multipliers?

  • Useful variables
  • Supply/demand difference, EWT, surge history
  • Model
  • Linear regression
  • Performance

31

slide-146
SLIDE 146

Can we predict surge multipliers?

  • Useful variables
  • Supply/demand difference, EWT, surge history
  • Model
  • Linear regression
  • Performance
  • R^2 scores range from 0.37 — 0.57

31

slide-147
SLIDE 147

Can we predict surge multipliers?

  • Useful variables
  • Supply/demand difference, EWT, surge history
  • Model
  • Linear regression
  • Performance
  • R^2 scores range from 0.37 — 0.57
  • Poor predicative power: missing variable (unfulfilled demand)

31

slide-148
SLIDE 148

Can we avoid surge pricing?

32

slide-149
SLIDE 149

Can we avoid surge pricing?

32

slide-150
SLIDE 150

Can we avoid surge pricing?

32

slide-151
SLIDE 151

Can we avoid surge pricing?

32

slide-152
SLIDE 152

Can we avoid surge pricing?

32

slide-153
SLIDE 153

Can we avoid surge pricing?

32

slide-154
SLIDE 154

Can we avoid surge pricing?

32

slide-155
SLIDE 155

Can we avoid surge pricing?

32 No Surge! EWT = 6min No Surge! EWT = 5min EWT = 2min

slide-156
SLIDE 156

Can we avoid surge pricing?

32 No Surge! EWT = 6min No Surge! EWT = 5min EWT = 2min

5 minutes 3 minutes 7 minutes

slide-157
SLIDE 157

Can we avoid surge pricing?

32 No Surge! EWT = 6min No Surge! EWT = 5min EWT = 2min

slide-158
SLIDE 158

Can we avoid surge pricing?

32 No Surge! EWT = 6min No Surge! EWT = 5min EWT = 2min

slide-159
SLIDE 159

Can we avoid surge pricing?

32 No Surge! EWT = 6min No Surge! EWT = 5min EWT = 2min

slide-160
SLIDE 160

Avoiding surge

33

slide-161
SLIDE 161

Avoiding surge

33

slide-162
SLIDE 162

Avoiding surge

33

slide-163
SLIDE 163

Avoiding surge

33

  • 10% — 15% of chances saving money
slide-164
SLIDE 164

Avoiding surge

33

  • 10% — 15% of chances saving money
  • On average, we can save 50% of prices
slide-165
SLIDE 165

Avoiding surge

33

  • 10% — 15% of chances saving money
  • On average, we can save 50% of prices
  • We built an app, but cannot release it
slide-166
SLIDE 166

What is the impact of surge pricing? Does it conform with Uber’s claim?

34

slide-167
SLIDE 167

What is the impact of surge pricing? Does it conform with Uber’s claim?

34

slide-168
SLIDE 168

What is the impact of surge pricing? Does it conform with Uber’s claim?

34

slide-169
SLIDE 169

What is the impact of surge pricing? Does it conform with Uber’s claim?

34

slide-170
SLIDE 170

What is the impact of surge pricing? Does it conform with Uber’s claim?

  • Supply

34

slide-171
SLIDE 171

What is the impact of surge pricing? Does it conform with Uber’s claim?

  • Supply
  • New: more new log-in cars in surging area

34

slide-172
SLIDE 172

What is the impact of surge pricing? Does it conform with Uber’s claim?

  • Supply
  • New: more new log-in cars in surging area
  • Move-in: more move-in cars from nearby areas

34

slide-173
SLIDE 173

What is the impact of surge pricing? Does it conform with Uber’s claim?

  • Supply
  • New: more new log-in cars in surging area
  • Move-in: more move-in cars from nearby areas
  • Move-out: less cars leaving the surging area

34

slide-174
SLIDE 174

What is the impact of surge pricing? Does it conform with Uber’s claim?

  • Supply
  • New: more new log-in cars in surging area
  • Move-in: more move-in cars from nearby areas
  • Move-out: less cars leaving the surging area
  • Demand

34

slide-175
SLIDE 175

What is the impact of surge pricing? Does it conform with Uber’s claim?

  • Supply
  • New: more new log-in cars in surging area
  • Move-in: more move-in cars from nearby areas
  • Move-out: less cars leaving the surging area
  • Demand
  • Booked: less cars getting booked in the surging area

34

slide-176
SLIDE 176

What is the impact of surge pricing? Does it conform with Uber’s claim?

  • Supply
  • New: more new log-in cars in surging area
  • Move-in: more move-in cars from nearby areas
  • Move-out: less cars leaving the surging area
  • Demand
  • Booked: less cars getting booked in the surging area
  • Old: more old cars staying the surging area

34

slide-177
SLIDE 177

Impact on supply and demand, in reality

State During surge Expected?

35

slide-178
SLIDE 178

Impact on supply and demand, in reality

State During surge Expected? New +2% Yes

35

slide-179
SLIDE 179

Impact on supply and demand, in reality

State During surge Expected? New +2% Yes Booked

  • 7%

Yes

35

slide-180
SLIDE 180

Impact on supply and demand, in reality

State During surge Expected? New +2% Yes Booked

  • 7%

Yes Old +14% Yes

35

slide-181
SLIDE 181

Impact on supply and demand, in reality

State During surge Expected? New +2% Yes Booked

  • 7%

Yes Old +14% Yes Move-in

  • 13%

No

35

slide-182
SLIDE 182

Impact on supply and demand, in reality

State During surge Expected? New +2% Yes Booked

  • 7%

Yes Old +14% Yes Move-in

  • 13%

No Move-out +14% No

35

slide-183
SLIDE 183

Summary of Uber study

36

slide-184
SLIDE 184

Summary of Uber study

  • Presented the first in-depth analysis of Uber
  • Covered dynamics of supply, demand, surge,

etc.

  • Detailed implementation of surge pricing

algorithm

36

slide-185
SLIDE 185

Summary of Uber study

  • Presented the first in-depth analysis of Uber
  • Covered dynamics of supply, demand, surge,

etc.

  • Detailed implementation of surge pricing

algorithm

  • Showed impact of surge on supply and demand

36

slide-186
SLIDE 186

Thanks!

  • Questions?