CityNet: A Multi-city Multi-modal Dataset for Smart City Applications
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
CityNet: A Multi-city Multi-modal Dataset for Smart
City Applications
Xu Geng1 , Yilun Jin1 , Zhengfei Zheng1 , Yu Yang2 , Yexin Li3 , Han Tian1 ,
Peibo Duan4 , Leye Wang5 , Jiannong Cao2 , Hai Yang1 , Qiang Yang1 , Kai Chen1
1
The Hong Kong University of Science and Technology, Hong Kong, China
arXiv:2106.15802v1 [cs.AI] 30 Jun 2021
2
The Hong Kong Polytechnic University, Hong Kong, China
3
JD Intelligent Cities Business Unit, Beijing, China
4
Northeastern University, China
5
Peking University, China
{xgeng, yilun.jin, zzhengak, yliby, htianab}@connect.ust.hk,
csyyang@comp.polyu.edu.hk, sakuragiduan@gmail.com,
leyewang@pku.edu.cn, jiannong.cao@polyu.edu.hk,
cehyang@ust.hk, {qyang, kaichen}@cse.ust.hk
Abstract
Data-driven approaches have been applied to many problems in urban computing.
However, in the research community, such approaches are commonly studied under
data from limited sources, and are thus unable to characterize the complexity
of urban data coming from multiple entities and the correlations among them.
Consequently, an inclusive and multifaceted dataset is necessary to facilitate more
extensive studies on urban computing. In this paper, we present CityNet, a multi-
modal urban dataset containing data from 7 cities, each of which coming from
3 data sources. We first present the generation process of CityNet as well as
its basic properties. In addition, to facilitate the use of CityNet, we carry out
extensive machine learning experiments, including spatio-temporal predictions,
transfer learning, and reinforcement learning. The experimental results not only
provide benchmarks for a wide range of tasks and methods, but also uncover
internal correlations among cities and tasks within CityNet that, with adequate
leverage, can improve performances on various tasks. With the benchmarking
results and the correlations uncovered, we believe that CityNet can contribute to
the field of urban computing by supporting research on many advanced topics.
1 Introduction
We have witnessed remarkable progress in advancing machine learning techniques (e.g., deep
learning [23, 9], transfer learning [15, 19], and reinforcement learning [17, 6]) for various smart city
applications [24] in recent years, such as predicting traffic speeds and flows, managing ride-hailing
fleets, forecasting urban environments, etc. However, due to the absence of a desirable open-source
dataset, most research works in urban computing investigate individual tasks using datasets from
limited sources. In fact, existing approaches can be further improved to deliver more intelligent
decisions by leveraging data from multiple sources. For example, while most research works on taxi
fleet management use taxi data only, in practice, such a task can benefit from additional data sources,
including real-time traffic speeds, weather conditions, POI distributions, demands and supplies of
other transports, etc. All these urban data are closely related to the taxi market from different aspects.
However, existing open datasets, such as PeMS [1], METR [4] and NYC Cabs [2] focus on individual
aspects of urban dynamics, e.g., traffic speeds or taxis. Furthermore, limited effort had been spent
35th Conference on Neural Information Processing Systems (NeurIPS 2021) Track on Datasets and Benchmarks.
Weather Citizens Traffic speed
CityNet Inflow (1)
Connectivity
Tasks
Outflow
Meteorology Service Context Beijing
POI
(2)
... ...
(3)
Taxi GPS
Service Context Hong Kong
Cities
Taxis
Idle-driving
City Layout Pick-ups
(c). Categories of sub-datasets
(a). Raw sources
(b). Sub-datasets and their correlations
Figure 1: Architecture of CityNet. Left: Three raw data sources of CityNet. Middle: Illustration of all
8 sub-datasets, whose sources are distinguished by underlines shown in Fig. 1(a), and categories are
distinguished by colors shown in Fig. 1(c). Right: Decomposition of the data dimensions into cities
and tasks. Directed curves indicate correlations to be discovered in this paper.
on unifying the spatio-temporal ranges of existing datasets for joint usage. For example, although
TaxiBJ [23], tdrive [21] and Q-traffic [10] all describe Beijing taxi, they are different in temporal
ranges and thus cannot be jointly utilized. Hence, an inclusive and spatio-temporally aligned dataset
is in urgent demand in urban computing to tackle above problems, so as to facilitate more accurate
algorithms and more insightful analyses.
In summary, there are two challenges that hamper building and utilizing such a dataset. First, in-
dividual urban datasets from various maintainers differ from each other in spatio-temporal ranges,
granularity, attributes, etc. This leads to difficulty in integrating them into unified formats under
aligned ranges and granularities for research purposes. Second, in addition to the dataset con-
struction, the correlations among sub-datasets must be uncovered so as to enhance performances by
identifying, sharing, and transferring related knowledge.
In this paper, we present CityNet, a dataset with data from multiple cities and sources for smart city
applications. Inspired by [16], we call CityNet multi-city multi-modal to describe the wide range of
cities and sources from which it is derived. Compared to existing datasets, CityNet has the following
features:
• Inclusiveness: As shown in Fig. 1(a), CityNet consists of 3 types of raw data (city layout, taxi,
meteorology) collected from 7 cities. Besides, we process the raw data into multiple sub-datasets
(Fig. 1(b)) to measure various aspects of urban phenomena. For example, we transform raw taxi
data into region-based measurements (taxi flows, pickups, and idle driving time) that show the
status of the transportation market and citizen activities.
• Spatio-temporal Alignment: To facilitate studies among cities and tasks, we use unified spatio-
temporal configuration, including geographical coordinates, temporal frames, and sampling
intervals to process all sub-datasets originating from the same city.
• Interrelatedness: We categorize sub-datasets into service and context data, as shown in Fig.
1(b) and (c), to demonstrate the status of urban service providers (e.g. taxi companies) and the
urban environment (e.g. weather) respectively. As depicted in Fig. 1(c), we verify the correlations
between sub-datasets, including the relationship (1) among services, (2) between cities and (3)
between contexts and services.
Leveraging the above features of CityNet, we conduct a series of experiments to benchmark various
popular machine learning tasks, including spatio-temporal prediction, transfer learning and rein-
forcement learning. The empirical studies reveal that utilizing such a multi-modal dataset leads to
performance improvements, and also point to the current challenges and future opportunities.
The contributions of this paper are summarized as follows:
• To the best of our knowledge, CityNet is the first multi-city multi-modal urban dataset to as-
semble and align sub-datasets from different tasks and cities. It is open to the research community
and can facilitate in-depth investigations on urban phenomenons from various disciplines.
• Extensive analysis and experiments on CityNet provide new insights for researchers. Upon
CityNet, we present diverse benchmarking results so as to motivate further studies in areas like
spatio-temporal predictions, transfer learning, reinforcement learning, and federated learning
in urban computing. Specifically, we use various tools to reveal and quantify the correlations
2
among different modalities and cities. We further demonstrate that appropriately utilizing the
mutual knowledge among correlated sub-datasets can significantly improve the performance of
various machine learning tasks.
CityNet has been open-sourced and is currently accessible at https://github.com/citynet-at-git/Citynet-
phase1. Besides, to empower the most advanced research in urban computing, we also plan to share
all resources related with this paper on TACC, a high-performance computing platform currently in
the initial stages of opening to the public. Please refer to Appendix A.7 for details.
2 Architecture of CityNet
In this section, we introduce the generation of CityNet, including the raw datasets, discretization,
generation of individual datasets and data integration across individual datasets.
2.1 Raw Data Collection
Taxi GPS points Taxi trajectories reflect rich information on citizen activities and the status of
the transportation network. We collect taxi GPS point records Gc in each city c, which consist of
trajectories of taxis Gc = {tra }a∈Ac , where Ac is the set of taxis in city c. Each trajectory tra is a
|tra |
sequence of tuples tra = [(posk , tk , occk )]k=1 , where posk denotes the position (i.e. longitude and
latitude) of the taxi at time tk , and occk denotes whether the taxi is occupied at time tk .
City Layout data We collect POI (points-of-interest), road network and traffic speed data to
describe the overall layout of the city. POI data (xpoi ) describe the functions served by each region. A
POI is defined as poi = (pospoi , catpoi ), where pospoi denotes the position (longitude and latitude) of
the POI, and catpoi denotes the category (e.g., catering, shopping) of the POI. We denote the set of all
POIs in city c as P OIc . The road network and traffic speed data (xspeed ) describe the real-time status
of the transportation system. A road segment is defined by its two endpoints as seg = (pos0 , pos1 ).
|sp |
Each segment is associated with its historical real-time traffic speeds spseg = [(tk , speedk )]k=1seg ,
where speedk denotes the speed of seg at time tk .
Meteorology data We collect weather data recorded at the airport of each city from rp51 every hour.
We process the weather data of each city into a matrix xmtr
c ∈ RTc /2×49 , where each 49-dimensional
vector contains information about temperature, air pressure, humidity, weather phenomena (e.g., rain,
fog), etc. We describe the detailed schema of the weather data in the Appendix A.3.
In total, we collect data for 7 cities: Beijing, Shanghai, Shenzhen, Chongqing, Xi’an2 , Chengdu2 and
Hong Kong. We provide data properties, including data range, size, and availability in Table 5, and
summarize notations used in this paper in Table 7 in the Appendix.
2.2 Raw Data Discretization
Discretization is the first pre-processing step that transforms raw datasets from continuous spatio-
temporal ranges to discrete space with fixed sampling intervals. This procedure enables building
indicators via aggregating discretized measurements from the same interval to quantify various
phenomena in cities. The formulations of the discretization in both spatial and temporal dimensions
are defined as follows:
Definition 1 (Region). Each city c is split into Wc × Hc square grids, each with size 1km×1km.
Each grid in a city is referred to as a region, denoted as r. We denote the set of all regions in city c
as Rc , and use rc (i, j) to denote the region in city c with coordinate (i, j).
Definition 2 (Timestamps). We define the set of available time stamps of city c as Tc = [τc − Tc +
1, ...τc ], where τc denotes the last timestamp, and Tc is the number of available time stamps of city c.
By default, we choose each time stamp to be a period of 30 minutes.
Definition 3 (Spatio-temporal Tensors). We denote a spatio-temporal tensor for city c (e.g. taxi flow
values) as xc ∈ RTc ×Wc ×Hc , and use xc (τ, i, j) and xc (τ ) to denote the value at region rc (i, j) and
the values for all regions in city c at timestamp τ , respectively.
1
https://rp5.ru/
2
Original data are retrieved from Didi Chuxing, https://gaia.didichuxing.com, under the Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International license. All data are fully anonymized.
3
2.3 Processing Individual Sub-datasets
Taxi Flow Dataset Region-wise human flows are important to traffic management and public
safety. Given taxi GPS points Gc , we extract inflow and outflow spatio-temporal tensors xin out
c , xc ∈
RTc ×Wc ×Hc as the number of taxis going into and out of each region within a timestamp [22]:
X X
xin
c (τ, i, j) = I (tk ∈ τ ∧ posk ∈ rc (i, j) ∧ posk−1 ∈
/ rc (i, j)) ,
tr∈Gc (posk ,tk ,occk )∈tr
X X (1)
xout
c (τ, i, j) = I (tk+1 ∈ τ ∧ posk ∈ rc (i, j) ∧ posk+1 ∈
/ rc (i, j)) ,
tr∈Gc (posk ,tk ,occk )∈tr
where I(X) is the indicator function. I(X) = 1 iff X is true.
Taxi Pickup and Idle Driving Dataset Demand and supply data of taxis are also important to
allocating transportation resources efficiently. Given taxi GPS points Gc , we derive the pickup and
idle driving spatio-temporal tensors xpickup
c , xidle
c to approximate the demands and supplies of taxis.
X X
xpickup
c (τ, i, j) = I(tk ∈ τ ∧ occk = 1 ∧ occk−1 = 0), (2)
tr∈Gc (posk ,tk ,occk )∈tr
1 X X
xidle
c (τ, i, j) = I(tk ∈ τ ∧ occk = 0)(tk+1 − tk ), (3)
|τ |
tr∈Gc (posk ,tk ,occk )∈tr
where |τ | denotes the length of a timestamp. Eqn. 2 computes the number of pickups of all taxis
at rc (i, j) at timestamp τ , which serves as a lower bound for taxi demands. Meanwhile, Eqn. 3
computes the total fraction of time in timestamp τ where taxis are running in idle.
POI Dataset For city c, we obtain POI data P OIc = {(pospoi , catpoi )} using AMap API 3 . Each
POI category4 is given an id from 0 to 13. We build POI tensors xpoi
c ∈ RWc ×Hc ×14 as follows:
X
xpoi
c (i, j, cat) = I(catpoi = cat ∧ pospoi ∈ rc (i, j)). (4)
(pospoi ,catpoi )∈P OIc
xpoi
c (i, j, cat) is the number of POIs with category cat in rc (i, j).
Road Connectivity Dataset We collect road map information, including highways and motorways
from OpenStreetMaps5 . According to the road maps, we build region-wise adjacency matrices
Aconn
c ∈ R|Rc |×|Rc | as:
X
Aconn
c (r0 , r1 ) = I(pos0 ∈ r0 ∧ pos1 ∈ r1 ), (5)
(pos0 ,pos1 )∈SEGc
where r0 , r1 ∈ Rc . Aconn
c (r0 , r1 ) represents the total number of roads connecting r0 and r1 , and
therefore, Aconn
c represents region-wise transportation connectivity within city c.
Traffic Speed Dataset The real-time speed data are processed into xspeedc ∈ RTc ×|SEGc | :
P
(tk ,speedk )∈sps I(tk ∈ τ ) · speedk
X
xspeed
c (τ, s) = P . (6)
s∈SEGc(tk ,speedk )∈sps I(tk ∈ τ )
It indicates the average traffic speed of the road segment s over timestamp τ .
As shown in Fig. 1(c), we categorize all taxi data (flow, pickup, and idle driving data) and traffic
speed data as service data as they represent the states of transport service providers, and the rest as
context data. We show statistics of all sub-datasets in Table 6 in the Appendix.
3
https://lbs.amap.com/
4
We maintain POIs with the following 14 categories: scenic spots, medical and health, domestic services,
residential area, finance, sports and leisure, culture and education, shopping, housing, governments and
organizations, corporations, catering, transportation, and public services.
5
https://openstreetmap.org/
4
2.4 Data integration
Service data and context data are usually integrated to jointly train machine learning models. In
CityNet, such a integration operation is simple using the ’JOIN’ operation on the aligned spatial or
temporal dimension. Using the taxi demand prediction problem as an example, we can represent the
integration of taxi pick-up data, POI data and meteorology data with the following relational algebra
formula: Π∗ (xpoi
c 1(i,j) xpickup
c 1τ xmtr
c ), which represents joining pick-up data with POI data on
the spatial dimension and joining pick-up data with meteorology data on the temporal dimension.
3 Data Mining and Analysis
In addition to data gathering and processing, we should also uncover and quantify the correlations
between sub-datasets in CityNet to provide insights on how to utilize the multi-modal data. In this
section, we use data mining tools to explore the correlations between service data and context data.
Inflow Outflow Pickup Idle-time Inflow Outflow Pickup Idle-time Inflow Outflow Pickup Idle-time
Beijing Shanghai Shenzhen
ARI 0.554 0.554 0.635 0.544 0.445 0.445 0.474 0.382 0.191 0.191 0.219 0.227
AMI 0.308 0.308 0.418 0.318 0.268 0.268 0.356 0.240 0.132 0.132 0.192 0.182
Chongqing Chengdu Xi’an
ARI 0.471 0.471 0.624 0.541 0.197 0.196 0.167 N/A 0.155 0.155 0.042 N/A
AMI 0.261 0.261 0.404 0.350 0.213 0.213 0.226 N/A 0.164 0.164 0.05 N/A
Table 1: ARI and AMI measurements between clustering assignments based on POI and those based
on taxi data. Both measurements range from [−1, 1]. Larger values indicate higher coincidence.
3.1 Correlation between POI and Taxi Data
Intuitively, taxi service data should be closely related to POIs. For example, the outflow of residential
areas should grow significantly in morning peak hours, while taxi pickups should be frequent in
business areas during evening rush hours. To confirm this, we cluster regions in city c to k = 3
disjoint clusters as: Cc,poi : Rc = Rc,1 ∪ Rc,2 ∪ · · · ∪ Rc,k , according to their POI vectors (xpoic )
using the k-means algorithm. Similarly, we also cluster regions using the average daily pattern of taxi
service data to obtain Cc,in , Cc,out , Cc,pickup , Cc,idle . We use the Adjusted Rand Index (ARI) and
the Adjusted Mutual Information (AMI) to measure the similarity between cluster assignments based
on POIs (Cc,poi ) and those based on taxi service data (Cc,in , etc). The results are shown in Table
1, where both ARI and AMI show positive correlations between both cluster assignments, which
confirms the relationship between the POI data and taxi service data.
Further more, we plot the average regional daily patterns of taxi service data from each POI-based
cluster in Beijing, Chengdu, and Xi’an in Fig. 2 in the Appendix. On one hand, in Fig. 2(a), taxi
patterns in Beijing are cohesive within each POI-based cluster and separable across clusters. On
the other hand, in Fig. 2(b), clusters with higher values significantly overlap in Xi’an and Chengdu,
which are cities with relatively low ARI and AMI in Table 1. This may be caused by the limited
number of regions in these cities. However, we can still identify two separable clusters in Fig. 2(b).
According to the results, we conclude that regions with similar POI distributions also share similar
taxi service patterns.
3.2 Correlation between Meteorology and Traffic Speed Data
Xi’an Chengdu
Rain -2.06∗∗ -0.64∗
Fog -0.78∗∗ -0.79∗∗
Table 2: Regression coefficients between weather phenomena and traffic speed in Chengdu and Xi’an.
∗ ∗∗
, indicate 5% and 1% significance under a double-sided t-test, respectively.
It is also intuitive that there exists correlations between weather and traffic speed. For example, when
confronted with rain or fog, drivers tend to drive with caution and thus slow down. Therefore, we
perform experimental studies to verify the correlations.
5
City Task Non Deep Learning Euclidean Deep Learning Graph Learning
RMSE/MAE ST-HA ST-LR ST-CNN MT-CNN ST-LSTM MT-LSTM ST-GCN MT-GCN ST-GAT MT-GAT
Inflow 7.90/2.97 6.07/2.59 5.60/2.35 5.45/2.39 5.67/2.31 5.48/2.40 5.48/2.27 5.34/2.19 5.59/2.43 5.36/2.20
Outflow 7.90/2.97 6.07/2.59 5.60/2.35 5.45/2.39 5.67/2.31 5.48/2.40 5.48/2.27 5.33/2.19 5.59/2.43 5.36/2.20
Beijing
Pickup 2.51/0.90 2.37/0.89 2.19/0.79 2.15/0.82 2.24/0.82 2.14/0.80 2.23/0.82 2.14/0.80 2.15/0.80 2.14/0.79
Idle 0.73/0.31 0.61/0.25 0.57/0.24 0.54/0.25 0.60/0.25 0.52/0.22 0.55/0.24 0.51/0.23 0.55/0.24 0.50/0.22
Inflow 19.41/7.93 11.72/5.93 8.29/4.36 8.41/4.51 8.60/4.57 8.67/4.60 10.40/5.26 9.77/5.05 10.05/5.11 9.64/4.99
Outflow 19.40/7.94 11.76/5.94 8.42/4.42 8.26/4.43 8.64/4.53 8.52/4.51 10.07/5.14 9.49/4.90 10.10/5.13 9.38/4.89
Shanghai
Pickup 2.24/0.88 1.94/0.84 1.66/0.75 1.64/0.75 1.66/0.74 1.61/0.73 1.78/0.79 1.72/0.80 1.75/0.77 1.71/0.76
Idle 1.23/0.54 0.89/0.41 0.86/0.44 0.75/0.41 0.85/0.41 0.73/0.36 0.86/0.42 0.79/0.38 0.81/0.40 0.78/0.38
Inflow 43.79/18.97 24.06/12.23 17.62/9.41 18.28/9.75 17.94/9.57 17.75/9.65 22.18/11.61 20.84/10.91 22.35/11.46 20.52/10.87
Outflow 43.85/18.97 24.08/12.23 18.08/9.54 18.45/9.53 17.70/9.54 17.80/9.64 22.08/11.33 20.82/10.81 21.99/11.28 20.59/10.82
Shenzhen
Pickup 6.78/2.48 4.91/2.09 4.29/1.83 4.24/1.76 4.00/1.75 3.99/1.76 4.39/2.02 4.24/1.86 4.50/1.92 4.21/1.84
Idle 4.25/1.59 2.51/1.09 2.29/0.98 2.16/1.00 1.88/0.88 1.90/0.93 2.33/1.05 2.17/0.97 2.29/1.01 2.18/0.98
Inflow 37.40/14.97 21.19/9.91 12.65/6.48 12.67/6.52 14.39/7.29 13.84/7.08 18.54/6.59 17.12/8.46 16.53/8.49 15.98/7.93
Outflow 37.36/14.87 21.23/9.85 12.90/6.60 12.99/6.60 14.30/7.10 13.68/6.95 18.75/8.87 16.91/8.29 18.12/9.12 15.94/7.93
Chongqing
Pickup 7.64/3.03 6.34/2.69 5.25/2.24 5.15/2.10 5.21/2.18 5.02/2.10 5.77/2.45 5.49/2.33 5.72/2.45 5.34/2.28
Idle 3.81/1.48 2.21/0.96 1.53/0.74 1.57/0.76 1.58/0.73 1.57/0.75 1.99/0.87 1.90/0.84 1.97/0.86 1.80/0.83
Inflow 87.00/55.28 43.69/29.72 27.79/19.20 25.77/17.85 28.13/19.39 26.33/17.94 34.77/23.86 31.70/21.77 31.62/21.67 30.67/21.59
Xi’an Outflow 86.66/55.29 43.58/29.59 27.59/18.97 25.32/17.55 27.90/18.19 25.92/17.69 35.24/24.15 31.04/21.49 29.88/22.18 28.96/20.12
Pickup 16.85/10.84 11.22/7.66 7.77/5.39 7.54/5.16 7.88/5.38 7.71/5.27 9.50/6.50 9.11/6.26 8.22/5.88 8.05/5.26
Inflow 117.5/71.1 58.92/37.72 31.22/21.09 30.02/20.40 34.50/23.57 32.78/22.42 45.10/29.49 41.27/27.62 40.02/27.27 38.47/25.84
Chengdu Outflow 116.9/71.5 58.98/37.88 31.91/21.46 29.74/20.20 33.55/22.86 32.77/22.22 43.66/28.47 40.39/27.10 35.59/24.72 35.20/24.58
Pickup 28.27/15.66 17.32/10.54 11.02/7.13 10.55/6.78 11.39/7.21 11.15/7.15 13.64/8.55 13.58/8.76 12.30/7.62 11.24/7.38
Table 3: Results for taxi prediction of all models and all tasks under 6 cities. ST-* and MT-* stands
for single-task and multi-task models respectively. The lowest RMSE/MAE in each task are bolded.
We perform both quantitative and qualitative studies. For quantitative studies, we perform least
squares regression on the average speed for each timestamp τ , i.e. MEANs xspeed c (τ, s) with the
weather phenomena xmtr c (τ ), and show the coefficients in Table 2. It can be shown that both rain
and fog have negative effects on the average traffic speed. We also plot the daily average speeds in
Xi’an, with and without rain or fog in Fig. 3 in the Appendix. We observe consistently higher average
speeds when there was no rain, and higher average speeds during day time when there was no fog.
4 Machine Learning Applications
In this section, we present experimental results of machine learning tasks that are supported by
CityNet, including spatio-temporal predictions (Section 4.1.1 and Appendix A.5.1), transfer learning
(Section 4.2), and reinforcement learning (Appendix A.5.2) to benchmark individual tasks. Also, the
benchmark results uncover the transferrability of knowledge across tasks and cities in CityNet.
4.1 Spatio-temporal Predictions
Spatio-temporal predictions are common and important tasks in smart city. Given the data in CityNet,
we provide benchmarks for two typical spatio-temporal prediction tasks, including the taxi service
prediction tasks in Section 4.1.1, and the traffic speed prediction task in Appendix A.5.1.
4.1.1 Taxi Service Prediction
We investigate the taxi service prediction task [22, 3], including flow and demand-supply prediction6 .
Problem Formulation We formulate the taxi service prediction problem as learning a single-step
prediction function fc that achieves the lowest prediction error over future timestamps:
x̃∗c (τ ) = f˜c ([x∗c (τ − L), ..., x∗c (τ − 1)]) ,
X
fc = arg min error (x̃∗c (τ ) , x∗c (τ )) . (7)
f˜c
τ ∈Tc
∗ = in, out, pickup, idle, joint,
where x∗c stands for the spatio-temporal tensor to predict, and error denotes some error function,
such as RMSE or MAE. We introduce two settings, the single-task and multi-task settings, where
x∗c denotes individual spatio-temporal tensors (e.g. in/outflow)
in out for the single-task
setting, and the
stacked spatio-temporal tensors xjoint
c = STACK xc , x c , x pickup
c , xidle
c for the multi-task
6
As stated in Sec. 2, we use pickup and idle driving as proxies for demand and supply.
6
setting. In other words, we achieve multi-task learning for multiple taxi service predictions through
direct weight sharing. We set the input as L = 5 closest temporal observations (i.e., 2.5 hours)7 .
Methods We study the following methods for taxi prediction.
• HA (History Average): Mean of historical values of the data.
• LR (Linear Regression): Ridge regression on historical values with a regularization of 0.01.
• CNN (Convolutional Neural Network): We implement our CNN based on STResNet [22]. We
use 6 residual blocks with 64 3 × 3 filters. Batch normalization (BN) is used.
• LSTM: We implement our LSTM based on ST-net [19], which first uses convolutions to model
spatial relations and uses LSTM to capture temporal dependencies. We use 3 residual blocks with
BN for convolutions, and use a single-layer LSTM with 256 hidden units.
• GCN (Graph Convolutional Network) [5] and GAT (Graph Attention Network) [14]: We apply
GCN and GAT with input features being historical values (single-task) or stacked historical
values of all tasks (multi-task) of input
data. For graph convolution, we take the adjacency matrix
as A = Aconn c + Apoi
c + Ac
neigh
/3 following [3], and use the normalized adjacency matrix
 = D−1/2 AD−1/2 for GCN.
Among them, HA, LR are traditional time series forecasting models, CNN, LSTM are deep learning
models on Euclidean structures (grids), while GCN, GAT are graph deep learning models utilizing
additional region-wise connections (e.g. POI and road connectivity).
Experimental Settings We split the earliest 70% data for training, the middle 15% for validation,
and the latest 15% for testing. All data are normalized to [0, 1] by min-max normalization and will be
recovered for evaluation. We use RMSE and MAE as metrics [20].
Table 3 shows the overall performances under both the single-task (ST-*) setting and the multi-task
(MT-*) setting for all investigated models. We make the following observations:
• Deep Learning or Not: Given sufficient data (e.g. 10 days, 1-2 months), deep learning models
including CNN, LSTM, GCN, and GAT outperform time-series forecasting methods like HA and
LR. This illustrates the power of deep learning in spatio-temporal predictions and the advantage
of leveraging correlations in both Euclidean and non-Euclidean spaces.
• Multi-task or Not: Among all 22 tasks, multi-task models achieve the lowest RMSE in 15
(68.2%) tasks, and the lowest MAE in 15 (68.2%) tasks. For each model, MT-* outperforms
ST-* in 80 (90.9%) cases in terms of RMSE and 66 (75%) cases in terms of MAE. We conclude
that taxi service predictions can be improved through a simple multi-task method, weight sharing,
which verifies the connections among heterogeneous taxi service data and paves way for future
research in multi-task spatio-temporal predictions.
• Graph Models or Not: Among all the prediction tasks, CNN-based models (CNN, LSTM)
achieve the best performances in most cases, while GNN-based models (GCN, GAT) work
well in Beijing only, which has the largest map size with the most complex intra-city relations.
We conclude that in cities with large map sizes, incorporating graphs to model region-wise
connections will improve prediction accuracy, while in cities with small map sizes, incorporating
graphs will hurt performance. One possible reason is that graphs built on a limited number of
regions tend to be dense, as shown in Table 6, which causes over-smoothing [7].
4.2 Inter-city Transfer Learning
As shown in Table 3, deep learning models generate accurate predictions when empowered with
abundant data. However, the level of digitization varies significantly among cities, and it is likely that
many cities cannot build accurate deep prediction models due to a lack of data. One solution to the
problem is transfer learning [12], which transfers knowledge from a source domain with abundant
data to a target domain without much data, and in our case, from one city to another. Therefore, we
7
For pickup and idle driving datasets, we aggregate 10-minute timestamps into 30-minute timestamps, and
use the aggregated tensors for prediction.
7
Full Data (Lower bound) 3-day Data (Upper bound) Fine-tuning/Source City RegionTrans/Source City
Target City Task
LR LSTM LR LSTM Beijing Shanghai Xi’an Beijing Shanghai Xi’an
Inflow 24.06/12.23 17.94/9.57 24.07/12.27 22.63/12.83 19.78/10.58 20.90/11.09 N/A 19.67/10.40 20.59/10.93 N/A
Outflow 24.08/12.23 17.80/9.64 24.09/12.27 22.75/12.87 19.82/10.58 20.95/11.09 N/A 19.62/10.49 20.70/10.95 N/A
Shenzhen
Pickup 4.91/2.09 3.99/1.76 4.92/2.09 4.52/2.07 4.25/1.86 4.37/1.92 N/A 4.21/1.82 4.31/1.89 N/A
Idle 2.51/1.09 1.90/0.94 2.51/1.09 2.39/1.17 2.12/0.97 2.16/1.10 N/A 2.12/0.97 2.13/1.00 N/A
Inflow 21.19/9.91 13.84/7.08 21.20/9.95 18.45/9.81 16.10/7.98 16.74/8.23 N/A 15.87/7.90 16.59/8.15 N/A
Outflow 21.24/9.85 13.68/6.95 21.24/9.86 18.22/9.69 15.85/7.83 16.61/8.13 N/A 15.72/7.83 16.46/8.08 N/A
Chongqing
Pickup 6.34/2.69 5.02/2.10 6.34/2.69 5.64/2.52 5.42/2.23 5.45/2.30 N/A 5.37/2.20 5.42/2.30 N/A
Idle 2.21/0.96 1.57/0.75 2.21/0.96 2.03/0.98 1.75/0.81 1.80/0.83 N/A 1.77/0.82 1.77/0.81 N/A
Inflow 58.92/37.72 32.79/22.42 59.29/38.45 48.39/33.12 43.49/28.88 46.30/30.08 45.24/29.69 43.43/28.60 45.74/29.92 44.43/29.51
Chengdu Outflow 58.98/37.88 33.55/22.86 59.37/38.61 47.77/32.69 43.12/28.44 46.06/30.12 44.66/29.35 42.73/28.66 45.59/29.86 43.69/29.17
Pickup 17.32/10.54 11.15/7.15 17.40/10.69 15.21/10.08 13.78/8.79 14.39/9.12 14.07/9.01 13.51/8.48 14.21/9.03 13.89/9.06
Table 4: Results for inter-city transfer learning from source cities (Beijing, Shanghai, Xi’an) to target
cities (Shenzhen, Chongqing, Chengdu). The lowest RMSE/MAE using limited target data are in
bold. Results under full and 3-day data are lower and upper bounds of errors of corresponding tasks.
carry out transfer learning experiments on CityNet to show that connections among cities can lead to
positive knowledge transfer, and to provide benchmarks for future works on urban transfer learning.
We investigate the problem of inter-city transfer learning on taxi service datasets. We formulate the
problem as to learn a function fcs ,ct that achieves the lowest prediction error on the target city ct ,
with the help of data from the source city cs :
x̃∗ct (τ ) = f˜cs ,ct x∗ct (τ − L), ..., x∗ct (τ − 1) , x∗cs ,
X
error x̃∗ct (τ ), x∗ct (τ ) .
fcs ,ct = arg min (8)
f˜ cs ,ct
τ ∈Tct
∗ = in, out, pickup, idle, joint, |x∗ct | |x∗cs |.
Experimental Settings We use LSTM, which is similar to ST-net in MetaST [19] as the base model,
and adopt the multi-task setting. For cities with large map sizes, we choose Beijing and Shanghai to
be source cities; for those with small map sizes, we choose Xi’an to be the source city, and the rest are
chosen as target cities. Xi’an is not used as the source city to transfer to Shenzhen and Chongqing due
to its much smaller map size. For all target cities, we choose a three-day, Thursday-Friday-Saturday
interval for training, and validation and test data remain the same.
Methods We study the following transfer learning methods:
• Fine-tuning: We train a model on the source city and then tune it on limited target data.
• RegionTrans [15]: RegionTrans computes a matching between source Rcs and target regions
Rct based on time-series (S-match) or auxiliary data (A-match) 8 correlation. The matching is
used as a regularizer, such that similar regions share similar features. In our experiments, we use
POI vectors to compute the matching, which belongs to A-match.
Results We show results of inter-city transfer learning in Table 4. We show results trained using
full and 3-day target data, which correspond to the lower and upper bound of errors, respectively.
We also show results of fine-tuning and RegionTrans. We make the following observations:
• Degradation under Data Scarcity. When only 3-day training data are available, LR, as a non-
deep learning model achieves similar performances as using full data, while LSTM suffers from
50% more error (see Chengdu). It indicates that compared to non-deep learning, deep learning
models show greater sensitivity to the amount of training data.
• Inter-city Correlations. Compared to using target data only (upper bound), transfer learning
achieves error reductions in all source-target pairs, the largest of which is about 15% (see
Shenzhen and Chongqing). The results indicate that there exist sufficient cross-city correlations
in CityNet, such that knowledge learned for one city can be beneficial to another.
• Impact of Context Data. Using POI data for region matching, RegionTrans achieves lower
error than fine-tuning in most cases, which verifies the connection between context and service
data, and underscores the importance of multi-modal data in CityNet.
• Domain Selection. We consistently achieve the best results when Beijing is used as the source
city, regardless of algorithms and the target city. One probable reason is that Beijing contains the
most regions, and hence the most complex regional service patterns.
8
In this paper, context data serve as auxiliary data.
84.3 Discussion
The spatio-temporal prediction experiments on the taxi service data have uncovered the correlation
among different tasks for individual sub-datasets in CityNet. Leveraging the mutual knowledge
among them, the prediction performances for individual tasks could be improved by simple multi-
task learning strategies like weight sharing. Also, via transfer learning, the cold start problem in
data-scarce cities could be alleviated by choosing a highly-related source domain and transfer the
abundant knowledge learned from it. This is supported by the cross-city correlations in CityNet.
To summarize, both the multi-task spatio-temporal prediction and the inter-city transfer learning
demonstrate the correlations among individual datasets in CityNet from different tasks and cities.
Thus, the inclusiveness and interrelatedness of CityNet are important features to facilitate the research
of various urban phenomena with new problem settings and new techniques.
5 Conclusion and Opportunities
We present a multi-city, multi-modal dataset for smart city named CityNet. To the best of our
knowledge, CityNet is the first dataset to incorporate urban data that are spatio-temporally aligned
from multiple cities and tasks. In this paper, to illustrate the importance of multi-modal urban data,
we show connections between service and context data using data mining tools, and provide extensive
experimental results on spatio-temporal predictions, transfer learning, and reinforcement learning
to show the connections among tasks and among cities. The results and the uncovered connections
endorse the potential of CityNet to serve as versatile benchmarks on various research topics, including
but not exclusive to:
• Transfer learning: In urban computing, knowledge required for different tasks, cities, or time
periods is likely to be correlated. Leveraging the transferable knowledge across these domains,
data scarcity problems in newly-built or under-developed cities can be alleviated. Containing
multi-city, multi-task data, CityNet is an ideal testbed for researchers to develop and evaluate
various transfer learning algorithms, including meta-learning [19], AutoML [8], and also transfer
learning on homogeneous or heterogeneous spatio-temporal graphs.
• Federated learning: Urban data are generated by various human activities and warehoused by
different stakeholders, including organizations, companies, and the government. Due to data
privacy regulations or the protection of commercial interests, collaborations between them should
be privacy-preserving. Federated learning (FL) [18] would become an effective solution enabling
such privacy-preserving multi-party collaboration. As CityNet contains data from multiple cities
and sources, it is appropriate to investigate various FL topics under various settings using CityNet,
with each party holding data from one source or one city.
• Explanation and diagnosis: Understanding social effects from data helps city governors make
wiser decisions on urban management. Therefore, besides to prediction and planning, we hope
that applications providing explicit insight into urban management, including causal inference
[11], uncertainty estimations, explainable machine learning can also be carried out on CityNet.
We hope CityNet can serve as triggers that lead to more extensive research efforts into machine
learning in smart city. For future work, we notice that CityNet is not yet multi-modal in terms of
transports, and thus plan to incorporate even more sources of data into CityNet, including subways
and buses.
9References
[1] Chao Chen. Freeway performance measurement system (pems). 2003.
[2] Nivan Ferreira, Jorge Poco, Huy T Vo, Juliana Freire, and Cláudio T Silva. Visual exploration
of big spatio-temporal urban data: A study of new york city taxi trips. IEEE transactions on
visualization and computer graphics, 19(12):2149–2158, 2013.
[3] Xu Geng, Yaguang Li, Leye Wang, Lingyu Zhang, Qiang Yang, Jieping Ye, and Yan Liu.
Spatiotemporal multi-graph convolution network for ride-hailing demand forecasting. In
Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 3656–3663,
2019.
[4] H.V. Jagadish, Johannes Gehrke, Alexandras Labrinidis, Yannls Papakonstantinou, Jignesh M.
Patel, Raghu Ramakrishnan, and Cyrus Shahabi. Big data and its technical challenges. Commu-
nications of the Acm, 57(7):86–94, 2014.
[5] Thomas N Kipf and Max Welling. Semi-supervised classification with graph convolutional
networks. arXiv preprint arXiv:1609.02907, 2016.
[6] Minne Li, Zhiwei Qin, Yan Jiao, Yaodong Yang, Jun Wang, Chenxi Wang, Guobin Wu, and
Jieping Ye. Efficient ridesharing order dispatching with mean field multi-agent reinforcement
learning. In The World Wide Web Conference, pages 983–994, 2019.
[7] Qimai Li, Zhichao Han, and Xiao-Ming Wu. Deeper insights into graph convolutional networks
for semi-supervised learning. In Proceedings of the AAAI Conference on Artificial Intelligence,
volume 32, 2018.
[8] Ting Li, Junbo Zhang, Kainan Bao, Yuxuan Liang, Yexin Li, and Yu Zheng. Autost: Efficient
neural architecture search for spatio-temporal prediction. In Proceedings of the 26th ACM
SIGKDD International Conference on Knowledge Discovery & Data Mining, pages 794–802,
2020.
[9] Yaguang Li, Rose Yu, Cyrus Shahabi, and Yan Liu. Diffusion convolutional recurrent neural
network: Data-driven traffic forecasting. arXiv preprint arXiv:1707.01926, 2017.
[10] Binbing Liao, Jingqing Zhang, Chao Wu, Douglas McIlwraith, Tong Chen, Shengwen Yang,
Yike Guo, and Fei Wu. Deep sequence learning with auxiliary information for traffic prediction.
In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery
and Data Mining. ACM, 2018.
[11] Wei Liu, Yu Zheng, Sanjay Chawla, Jing Yuan, and Xie Xing. Discovering spatio-temporal
causal interactions in traffic data streams. In Proceedings of the 17th ACM SIGKDD interna-
tional conference on Knowledge discovery and data mining, pages 1010–1018, 2011.
[12] Sinno Jialin Pan and Qiang Yang. A survey on transfer learning. IEEE Transactions on
knowledge and data engineering, 22(10):1345–1359, 2009.
[13] Xiaocheng Tang, Zhiwei (Tony) Qin, Fan Zhang, Zhaodong Wang, Zhe Xu, Yintai Ma, Hongtu
Zhu, and Jieping Ye. A deep value-network based approach for multi-driver order dispatching.
In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery
and Data Mining, pages 1780–1790, 2019.
[14] Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua
Bengio. Graph attention networks. arXiv preprint arXiv:1710.10903, 2017.
[15] Leye Wang, Xu Geng, Xiaojuan Ma, Feng Liu, and Qiang Yang. Cross-city transfer learning
for deep spatio-temporal prediction. In Proceedings of the 28th International Joint Conference
on Artificial Intelligence, pages 1893–1899. AAAI Press, 2019.
[16] Ying Wei, Yu Zheng, and Qiang Yang. Transfer knowledge between cities. In Proceedings of
the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining,
pages 1905–1914, 2016.
[17] Zhe Xu, Zhixin Li, Qingwen Guan, Dingshui Zhang, Qiang Li, Junxiao Nan, Chunyang Liu,
Wei Bian, and Jieping Ye. Large-scale order dispatch in on-demand ride-hailing platforms:
A learning and planning approach. In Proceedings of the 24th ACM SIGKDD International
Conference on Knowledge Discovery and Data Mining, pages 905–913, 2018.
10[18] Qiang Yang, Yang Liu, Tianjian Chen, and Yongxin Tong. Federated machine learning: Concept
and applications. ACM Transactions on Intelligent Systems and Technology (TIST), 10(2):1–19,
2019.
[19] Huaxiu Yao, Yiding Liu, Ying Wei, Xianfeng Tang, and Zhenhui Li. Learning from multiple
cities: A meta-learning approach for spatial-temporal prediction. In The World Wide Web
Conference, pages 2181–2191, 2019.
[20] Bing Yu, Haoteng Yin, and Zhanxing Zhu. Spatio-temporal graph convolutional networks: a
deep learning framework for traffic forecasting. In Proceedings of the 27th International Joint
Conference on Artificial Intelligence, pages 3634–3640, 2018.
[21] Jing Yuan, Yu Zheng, Xing Xie, and Guangzhong Sun. Driving with knowledge from the physi-
cal world. In Proceedings of the 17th ACM SIGKDD international conference on Knowledge
discovery and data mining, pages 316–324, 2011.
[22] Junbo Zhang, Yu Zheng, and Dekang Qi. Deep spatio-temporal residual networks for citywide
crowd flows prediction. In Proceedings of the AAAI Conference on Artificial Intelligence,
volume 31, 2017.
[23] Junbo Zhang, Yu Zheng, Dekang Qi, Ruiyuan Li, and Xiuwen Yi. Dnn-based prediction
model for spatio-temporal data. In Proceedings of the 24th ACM SIGSPATIAL International
Conference on Advances in Geographic Information Systems, pages 1–4, 2016.
[24] Yu Zheng, Licia Capra, Ouri Wolfson, and Hai Yang. Urban computing: concepts, method-
ologies, and applications. ACM Transactions on Intelligent Systems and Technology (TIST),
5(3):1–55, 2014.
[25] Ming Zhou, Jiarui Jin, Weinan Zhang, Zhiwei Qin, Yan Jiao, Chenxi Wang, Guobin Wu, Yong
Yu, and Jieping Ye. Multi-agent reinforcement learning for order-dispatching via order-vehicle
distribution matching. In Proceedings of the 28th ACM International Conference on Information
and Knowledge Management, pages 2645–2653, 2019.
Acknowledgments and Disclosure of Funding
This project is sponsored by RGC Theme-based Research Scheme (No.T41-603/20R). The raw
datasets are retrieved from Didi Chuxing GAIA Initiative (https://gaia.didichuxing.com) and some
other open-source channels. We appreciate the effort from Hao Liu, Kaiqiang Xu and Xiaoyu Hu
({liuh,kxuar,huxiaoyu}@ust.hk, HKUST) throughout this research work.
Checklist
1. For all authors...
(a) Do the main claims made in the abstract and introduction accurately reflect the paper’s
contributions and scope? [Yes]
(b) Did you describe the limitations of your work? [Yes]
(c) Did you discuss any potential negative societal impacts of your work? [No] We only
publish the discretized and aggregated measurements of our dataset, which will not
reveal sensitive information of individual users.
(d) Have you read the ethics review guidelines and ensured that your paper conforms to
them? [Yes]
2. If you are including theoretical results...
(a) Did you state the full set of assumptions of all theoretical results? [N/A]
(b) Did you include complete proofs of all theoretical results? [N/A]
3. If you ran experiments (e.g. for benchmarks)...
(a) Did you include the code, data, and instructions needed to reproduce the main experi-
mental results (either in the supplemental material or as a URL)? [Yes]
(b) Did you specify all the training details (e.g., data splits, hyperparameters, how they
were chosen)? [Yes]
11(c) Did you report error bars (e.g., with respect to the random seed after running experi-
ments multiple times)? [No]
(d) Did you include the total amount of compute and the type of resources used (e.g., type
of GPUs, internal cluster, or cloud provider)? [Yes]
4. If you are using existing assets (e.g., code, data, models) or curating/releasing new assets...
(a) If your work uses existing assets, did you cite the creators? [Yes]
(b) Did you mention the license of the assets? [Yes]
(c) Did you include any new assets either in the supplemental material or as a URL? [No]
(d) Did you discuss whether and how consent was obtained from people whose data you’re
using/curating? [Yes]
(e) Did you discuss whether the data you are using/curating contains personally identifiable
information or offensive content? [Yes]
5. If you used crowdsourcing or conducted research with human subjects...
(a) Did you include the full text of instructions given to participants and screenshots, if
applicable? [N/A]
(b) Did you describe any potential participant risks, with links to Institutional Review
Board (IRB) approvals, if applicable? [N/A]
(c) Did you include the estimated hourly wage paid to participants and the total amount
spent on participant compensation? [N/A]
12A Supplementary Materials
A.1 Overview of CityNet data
In this section, we provide the overview of CityNet in Table 5 and some basic statistical properties in
Table 6 as a supplement to the introduction section.
Time Range City Layout Taxi
City Meteorology
(inclusive) Longitude range (E) Latitude range (N) POI Road Speed In/Outflow Pickup Idle
Beijing 10/20 - 10/29, 2013 115.9045-116.9565 39.5786-40.4032 3 3 7 3 3 3 3
Shanghai 08/01 - 08/31, 2016 121.2054-121.8176 30.9147-31.4202 3 3 7 3 3 3 3
Shenzhen 12/01 - 12/31, 2015 113.8097-114.3087 22.5085-22.7739 3 3 7 3 3 3 3
Chongqing 08/01 - 08/31, 2017 106.3422-106.6486 29.3795-29.8322 3 3 7 3 3 3 3
Xi‘an 10/01 - 11/30, 2016 108.9219-109.0093 34.2049-34.2799 3 3 3 3 3 7 3
Chengdu 10/01 - 11/30, 2016 104.0421-104.1296 30.6528-30.7278 3 3 3 3 3 7 3
Hong Kong 10/13 - 12/22, 2020 113.9878-114.2459 22.2745-22.4670 7 3 3 7 3
Table 5: Overview of data in CityNet, including spatio-temporal ranges and availability of sub-
datasets in each city. Cities in CityNet have diverse properties. Beijing has the largest map size
with |Rc | ≈ 10000. Shanghai, Shenzhen and Chongqing have large maps with |Rc | > 1000, while
Chengdu and Xi’an having small maps with |Rc | < 100.
In/Outflow Pickup Idle-time In/Outflow Pickup Idle-time In/Outflow Pickup Idle-time
Taxi Service
Beijing Shanghai Shenzhen
Shape (480,106,92) (1440,106,92) (1448,62,57) (4464,62,57) (1448,50,30) (4464,50,30)
Mean 6.682 0.389 0.488 27.181 0.637 1.618 88.244 2.893 5.488
Max 671.0 340.0 243.3 1440.0 304.0 438.0 3601.0 302.0 583.7
Sparsity 0.695 0.920 0.788 0.406 0.838 0.537 0.313 0.715 0.412
Chongqing Chengdu Xi’an
Shape (1488,31,51) (4464,31,51) (2928,9,9) (8784,9,9) - (2928,9,9) (8784,9,9) -
Mean 54.761 2.651 3.944 343.309 21.835 - 227.624 12.897 -
Max 1572.0 495.0 411.8 3646.0 554.0 - 2768.0 250.0 -
Sparsity 0.411 0.727 0.477 0.028 0.102 - 0.001 0.111 -
Road Connectivity Beijing Shanghai Shenzhen Chongqing Chengdu Xi’an Speed Chengdu Xi’an Hong Kong
# roads 180,488 151,124 69,612 31,138 6,846 11,810 # roads 1,676 792 618
Max 65 67 32 27 35 23 Max 170.6 149.22 111.0
Mean (×1e-4) 20.07 95.35 180.33 122.69 6752.02 4429.20 Mean 31.93 32.49 56.09
Sparsity 0.999 0.997 0.994 0.994 0.823 0.838 Sparsity 0.047 0.053 0.027
Table 6: Statistics of Taxi Service, Road Connectivity and Traffic Speed sub-datasets within all cities.
The temporal granularity is 30 minutes for in/outflow data, and 10 minutes for pickup, idle driving
and traffic speed data. Sparsity denotes the fraction of 0s within all entries of the sub-dataset.
A.2 Summary of Notations
We summarize all notations used in this paper in Table 7.
A.3 Details of Meteorology Dataset
Given the weather data matrix xmtr
c ∈ RTc /2×49 , each dimension in the 49-dimensional vector
mtr
xc (τ ) has the following meaning:
• 0: Air temperature (degrees Celsius);
• 1: Atmospheric pressure (mmHg);
• 2: Atmospheric pressure reduced to mean sea level (mmHg);
• 3: Relative humidity (%);
• 4: Mean wind speed (m/s);
• 5: Horizontal visibility (km);
• 6: Dewpoint temperature (degrees Celcius);
• 7-24: Multi-hot encodings of whether phenomena. Each index represents fog, patches fog,
partial fog, mist, haze, widespread dust, light drizzle, rain, light rain, shower, light shower,
13Notation Meaning
c A specific city, e.g. Beijing, Shanghai
Wc , Hc Width and height of the city c in grids
r A region, i.e. a 1km×1km grid
rc (i, j) The region in city c with grid coordinate (i, j)
Rc The set of all regions in city c
τ A timestamp, a 30-minute interval by default
τc The last valid timestamp of city c
Tc The set of valid timestamps of city c
Tc # valid timestamps of city c
Gc The set of all taxi GPS points in city c
Ac The set of all taxis in city c
tra The trajectory of taxi a
(pos, t, occ) A taxi GPS point
P OIc The set of POIs in city c
(pospoi , catpoi ) The position and category of a POI
xc A spatio-temporal tensor for city c
xc (i, j, τ ) The value at region rc (i, j) at τ
xc (τ ) The spatio-temporal values of Rc at time τ
seg = (pos0 , pos1 ) A road segment and its two endpoints
SEGc The set of road segments in city c
spseg The sequence of history traffic speeds of seg
Table 7: Summary of Notations
(a) Beijing (b) Chengdu and Xi’an
Figure 2: Average daily pattern of taxi service data in (a) Beijing, (b) Chengdu and Xi’an. We
aggregate all values at each timestamp from all days, and represent mean values at each timestamp
using solid lines, and standard deviations using shades.
in the vicinity shower, heavy shower, thunderstorm, light thunderstorm, in the vicinity
thunderstorm, heavy thunderstorm, and no special weather phenomena, respectively;
• 25-42: One-hot encoding of wind direction. Each index represents no wind, variable
direction, E, E-NE, E-SE, W, W-NW, W-SW, S, S-SE, S-SW, SE, SW, N, N-NE, N-NW, NE,
and NW, respectively;
• 43-48: One-hot encoding of wind covering. Each index represents no significant clouds, cu-
mulonimbus clouds, few clouds, scattered clouds, broken clouds, and overcast, respectively.
A.4 Visualization for Data Mining and Analysis
We provide the visualization for data mining analysis in Section 3 in Figure 2 and 3.
14Daily Avg speed at each hour in Xi'an Daily Avg speed at each hour in Xi'an
40 40
38 38
36 36
Speed (km/h)
Speed (km/h)
34 34
32 32
30 30
28 28
26 Rain 26 Fog
24 No Rain 24 No Fog
0:00 4:00 8:00 12:0016:0020:0024:00 0:00 4:00 8:00 12:0016:0020:0024:00
Time Time
(a) Rain (b) Fog
Figure 3: Daily average speed in Xi’an, with and without rain or fog. Shades represent half of the
standard deviation.
RMSE/MAE Xi’an Chengdu Hong Kong
HA 8.7475/5.0312 7.9507/4.5202 6.784/3.8175
LR 8.1432/4.5687 7.4246/4.1844 5.7752/3.3942
GCN 7.5872/4.1807 7.0598/3.9894 5.5043/3.1405
GAT 7.5217/4.0987 6.9622/3.8825 5.5722/3.1465
Table 8: Results for traffic speed prediction in Xi’an, Chengdu, and Hong Kong.
A.5 More Machine Learning Benchmarks
A.5.1 Traffic Speed Prediction
We investigate the traffic speed prediction problem [20, 9]. We formulate the problem as learning a
multi-step prediction function fc (·):
[x̃sc (τ ), ..., x̃sc (τ + Lo − 1)] = f˜c ([xsc (τ − Li ), ..., xsc (τ − 1)]) ,
(9)
X
fc = arg min error (x̃sc (τ ) , xsc (τ )) ,
f˜c
τ ∈Tc
where xsc is a shorthand for the traffic speed sub-dataset xspeed
c . In this experiment, the input length
Li is set to 12 (2 hours) and the output length Lo is set to 6 (1 hour). We investigate four methods, HA,
LR, GCN and GAT described in Sec. 4.1.1. For i, j ∈ SEGc , i = (posi0 , posi1 ), j = (posj0 , posj1 ),
the adjacency matrix A for GNN models is calculated as
1, posi0 = posj1 or posi1 = posj0 .
A(i, j) = (10)
0, otherwise
Table 8 shows the results of the traffic speed prediction tasks in three cities for all examined models.
A.5.2 Reinforcement Learning
Planning for adequate travel resource allocation is important in smart city. Based on the data in
CityNet, we provide benchmarks for the taxi dispatching task, where operators continually dispatch
available taxis to waiting passengers in real-time so as to maximize the long-term total revenue of the
taxi system.
Problem Statement. Given historical requests from passengers, we learn a real-time taxi dispatching
policy. At every timestamp τ , we dispatch available taxis to the current passengers according to the
policy so as to maximize the total revenue of all taxis in the long run. Specifically, we divide the city
into uniform hexagon grids, instead of square ones hexagon as per previous studies [13, 17].
Methods. We study the following methods for taxi dispatching.
• Greedy is a heuristic algorithm that dispatches the closest available taxi to each passenger
one by one.
15Completion rate Accumulated revenue
Date
Greedy LLD LPA Greedy LLD LPA
Oct. 1st 0.689 0.734 0.718 1,017,144 1,035,724 1,016,410
Oct. 2nd 0.705 0.765 0.752 969,340 984,424 975,157
Oct. 3rd 0.680 0.734 0.747 943,422 961,263 969,341
Oct. 4th 0.695 0.810 0.830 965,896 1,034,539 1,041,978
Oct. 5th 0.708 0.782 0.811 984,225 1,021,727 1,032,165
Oct. 6th 0.735 0.832 0.825 979,356 1,028,292 1,020,174
Oct. 7th 0.774 0.872 0.889 932,641 977,996 980,397
Table 9: Results for taxi dispatching in Chengdu
• LLD is an optimization based approach formulated as Eqn. 11, where aij = 1 if taxi j is
dispatched to passenger i and 0 otherwise; A(i, j) is the immediate revenue to taxi j after
driving passenger i to his/her destination. The definition of the immediate revenue follows
[17].
m X
X n
max A(i, j)aij
aij
i=0 j=0
Pm (11)
i=0 aij = 1, j = 1, 2, 3, ..., n
s.t.
Pn aij = 1, i = 1, 2, 3, ..., m
j=0
• LPA [17] is a reinforcement learning based planning approach. We first adopt SARSA [17]
to learn the expected long-term revenue of each grid in each period. Then, based on these
expected revenues, we dispatch taxis to passengers by the same optimization formulation
as Eqn. 11, except that we replace A(i, j) with the scores learned by SARSA. LPA tries to
maximize the total revenue of the system in the long-run instead of immediate revenues.
Experimental Settings Based on historical request data, we implement a simulator according to
[17, 25] to simulate how the taxi system operates, and evaluate these methods on our simulator.
We show taxi dispatching results in Chengdu in Table 9, where Completion rate denotes the ratio of
completed requests within all requests, and Accumulated revenue denotes the total revenue received
by all taxis in the whole day. We draw the following conclusions from the experiment results:
• Greedy algorithm, which does not consider any global optimization targets, performs the
worst compared to LLD and LPA. Taking global optimization targets into consideration
leads to an improvement of 5%-20% in completion rate and 2%-8% in revenue.
• LPA performs better than LLD most of the time. Since LPA optimizes the expected long-
term revenues at each dispatching round, while LLD only focuses on the immediate reward,
LPA should compare favorably against LLD.
A.6 Details of Machine Learning Experiments
We provide detailed settings for our machine learning experiments. Note that we only tune these
settings to make sure all models converge, rather than to achieve state-of-the-art results.
A.6.1 Taxi Service Prediction
For CNN and LSTM, we train models with learning rate 1e-4, weight decay 0 using Adam optimizer
for 75 epochs, and choose the model with the lowest validation error for testing. We set the batch
size as 64 for CNN, and use variable batch sizes (Beijing 4, Shanghai 8, Shenzhen/Chongqing 16,
Xi’an/Chengdu 32) for LSTM. Weather data are used in training the model following the method in
[22]. Each result is an average of 5 independent runs.
16A.6.2 Inter-city Transfer Learning
For fine-tuning, we tune the model trained over the source city with learning rate 5e-5, weight decay
0 using Adam optimizer for 100 epochs, and choose the model with the lowest validation error on the
target city for testing. For RegionTrans, the weight for feature consistency is taken as 0.01.
A.7 Hardware Configurations and Data-sharing Approaches
All experiments included in this paper are conducted with Python 3.6 and PyTorch 1.6.0 environment
on Tesla V100 GPUs. The hardware is affiliated to the Turing AI Computing Cloud (TACC)9
maintained by the iSING Lab at HKUST10 .
To empower smart city applications including transportation, healthcare, finance, etc., we are building
TACC, an open platform that provides high performance computing resources, massive smart city
data including our CityNet, and state-of-the-art machine learning algorithmic paradigms. Researchers
can train their own machine learning models on CityNet on TACC equipped with GPUs and RDMA
for free. We plan to open the platform to the the academic community in Hong Kong by December
2021, and to the global researchers by June 2022.
A.8 Author statement and license
We bear all responsibility in case of violation of rights during using, processing and distributing these
data. This work is licensed under a CC BY 4.0 license.
9
https://turing.ust.hk/tacc.html
10
https://ising.cse.ust.hk/
17You can also read
