International Journal of Advanced Engineering Research and Science (IJAERS) [Vol -6, Issue-6, June- 2019]

https://dx.doi.org/10.22161/ijaers.6.6.49 ISSN: 2349-6495(P) | 2456-1908(O)

Data Collection and Prediction of Urban

Transport Flow using Neural Networks
Luis Felipe Hau, José Claudio Vianna Junior, Patrick Ribeiro, Verônica
Isabela Quandt
Department of computer engineering, Positivo University, Curitiba - Paraná, Brazil

Abstract—Smart cities can use artificial neural networks This set of information could be used to create an
to provide more accurate information about public artificial neural network that analyzes all th is data and
transportation schedules, and thus help the population tries to find a possible connection between them, so that it
plan their day to day activities. In this context, this paper creates an algorith m to predict situations of delay or
describes the essential steps for the acquisition and advance, becoming a tool to help p rofess ionals in the area
processing of data, and the creation of a neural network of data analysis and even to the user of the bus network.
model capable of predicting possible delays or advances
In the scenario of the bus lines of public transport, one of
on bus lines in the city of Curitiba, Paraná. The neural
the known issues is the compliance with the established
network considers traffic data, climate, time and history
schedules. Because it is a problem that in many cases is
of a public transport line. The article details all phases of
caused by factors that can not be controlled, it is not
collection and treatment, as well as how informatio n is
always possible to prevent it from happening. Predict
inserted into the network and what are the obtained
these delays allows interested parties to have this
results.
informat ion in advance and can decide how to work
Keywords—Neural networks, transport prediction, around the situation [3].
smart cities.
With population growth also increase the challenges for
I.INTRODUCTION government, business and academia [4]. The analysis of
data to create resources for intelligent cities has been the
Collective public t ransportation is of g reat impo rtance to
subject of several studies in both the academic and
Brazilian cit ies. These transit systems provide invaluable
business environments. This technique of collecting and
access and locomotion to some of the country's poorest
processing data can be of great value to companies and
citizens. Furthermore, collective public t ransportation
users who could benefit fro m a great amount of
vehicles help to reduce traffic congestion, mitigate
informat ion, planning and improving their activ ities, but
emissions from indiv idual automobiles and contribute to
also to the government that could benefit from the
an overall strategy to promote cleaner and
improvement in the service provided. Res earch shows
environmentally-friendly cities.
that the greatest cause of dissatisfaction among the
The urban public transport plays an important role in the Brazilian population with public transportation are the
current configuration of urban displacement as a means of problems with capillarity and frequency, slowness and
transport that provides the interconnection between the frequent delays, which, according to the research [5],
various regions of the cities. It is an alternative to the cause the population to us e less public transportation.
reduction of serious problems found in cit ies such as:
According to [6], congestion concerns all individuals.
congestion, traffic accidents and environmental impacts
Brazilian metropolitan areas live a n ightmare d ifficu lt to
[1].
measure, wh ich are urban congestion. The feeling o f
Forecasting public transport delays can be an optimized wasted time in front of a huge congestion is worrying,
tool that drivers and passengers could use to plan their and there are few people who know how to live with this
daily tasks. This prediction can be obtained by analyzing reality naturally. In recent years, millions of people have
data directly or indirectly linked to the line punctuality lost money and time because of congestion [7] and there
situation. Data collection is an important aspect of urban is a considerable increase in the p rice of car t rips during
computing and is a determining factor in build ing smart congestion [8, 9].
cities [2].

www.ijaers.com Page | 476

International Journal of Advanced Engineering Research and Science (IJAERS) [Vol -6, Issue-6, June- 2019]
https://dx.doi.org/10.22161/ijaers.6.6.49 ISSN: 2349-6495(P) | 2456-1908(O)
Knowing this reality this article proposes a solution of data were collected for 23 days. To evaluate the
public utility so that the population knows of the expected punctuality of the bus, it was admitted that the process of
delays of a bus line. The solution brings advantages such generating a random sequence of delays, anticipations and
as reduction of the issues involved, and can benefit users certain hours on a given day must be influenced by what
of transport, service provider co mpanies and government happened in the previous days and can therefore be
entities. described by a Markov chain. Besides the estimation,
some simu lated samples of these chains were also used in
This article is organized as follo ws. Section II discusses
the work. Ho wever, it was verified that for samples of
related work. Section III describes the data collection
size up to 50 days, there is not sufficient informat ion to
considered in this paper, examines the data model and
detect a dependency structure, even if the practicality of
how they may or may not directly in fluence the final
the use of modeling of a variable through the chain was
result. Section IV describes the experiments performed
evidenced.
and how the data obtained was added to the neural
network and which tool was used for this, also shows the The idea of improv ing the information offered to public
results obtained and the network responses. Section V transport users, based on informat ion provided by other
concludes the paper and presents future work. individuals, was worked by Lucio [12]. In the wo rk,
collective intelligence is used, which is described as a
II. RELATED WORK
form of distributed intelligence, constantly improved by
The attempt to pred ict possible delays of collective public its users and coordinated in real time, resulting in the
transport vehicles has already been considered in other creation of knowledge through collaboration. The work
works. So me of these studies involve computational shows how the resources provided by mobility in
intelligence, but the vast majority use only historical data conjunction with collective intelligence can be used to
and some other technique. On the other hand, some create Intelligent Transport Systems (ITS). In this
studies have more similar characteristics with this work scenario, the data required for the creation of these
and also use weather and traffic data. intelligent systems are provided by the users of the public
In the work of Maciel [3], the performance of regression transportation through their mobile devices, providing the
algorith ms in historical data for the fo recast of the start construction of a large collection of informat ion of the
and end time of day t rips is evaluated. The main idea was transportation system through the contribution of the
to evaluate the performance of regression algorith ms and users.
with them it was verified that, for both the start and end
III. EXPERIMENTAL ANALYSIS
time of the trip, the med ian of the errors was
For the pred iction times of the path of a line of public
approximately 28 and -167 seconds respectively. The
transport in Curitiba, it is necessary to collect data about
work shows that the quality of the forecasts also changes
this line and additional data that may influence its path.
over the course of the week, where the worst results were
These lines have routes that meander through the city,
obtained on Monday and the best on Wednesday and
being of great use by the passengers, passing through
Thursday. The same behavior o f the days of the week was
terminals and streets that have a great flow of people and
verified in the hours of the day, where the start and end
vehicles. The additional data is based on climate
times of the usual Brazilian work schedule obtained the
informat ion and traffic incidents in a generic way and
most inconsistent results. They also considered some
without any specific category, which, for both, has a great
climate data and their influences.
impact on the flow of vehicles. The acquisition process is
In the work of Moraes Filho [10], a pro ject called to: identify lines, collect real-t ime basis and time
CittaMobi is presented, which is a set of solutions that schedules of the line and of each vehicle; identify and
aims to make public t ransport information available to collect weather data fro m the region of Curitiba, targeted
bus users. The application provides real-time pred ictions in temperature, humidity, wind speed and description of
of the arrival of the bus, the locations of the closest the weather; and, finally, co llect traffic data based on line
points, together with the lines that pass through them, and locations.
some details related to each bus, for examp le, if the bus is
After the data is collected, the stored information is
adapted to the holders of special needs, or not.
analyzed. Before submitting them for p rocessing with the
On the other hand, the work of Serafim [11] consists of an neural network, it is necessary to identify and translate the
experiment carried out in the area of public transportation, informat ion. So me of the data collected is based on
with data obtained by an observer and collected with natural language. It is not necessary to carry out a more
direct observation of the arrival of the bus. In the study, advanced classification based on Natural Language

www.ijaers.com Page | 477

International Journal of Advanced Engineering Research and Science (IJAERS) [Vol -6, Issue-6, June- 2019]
https://dx.doi.org/10.22161/ijaers.6.6.49 ISSN: 2349-6495(P) | 2456-1908(O)
Processing (NLP), since the terms used are very limited. With this in mind, so me data on climat ic conditions in the
This classification is necessary only for the weather and region of circulation of the bus are necessary. Climate
traffic data, to simp lify data entry to a model of the neural informat ion was found in the form of web service. The
network. web service chosen for use in this research was the HG
Weather - Weather Forecasting API [16] which is a
3.1 Data Collection
project designed to disseminate in formation for free. In
To perform the predict ion, historical data are needed, this web service, we can obtain data using: the city code
especially those that may have a connection with public (WOEID), Geo IP, Geolocation or by the name of the
transport bus delays. These data were collected in several city.
ways and in several formats: weather data, history of
The mode chosen was using the WOEID code wh ich,
delays of the chosen line, data of traffic and traffic flo w in
according to [17], is an acronym for Where On Earth
the region transited by the bus at the time of collection,
Identifier, wh ich marks the location o f cities and
among others.
identifies each with a specific code. For the city of
3.1.1 Climatic Data Collection Curitiba the WOEID is 455822, and this way o f obtaining
data was chosen because it does not require an access key.
Authors report that the adverse weather conditions cause
significant changes in travel decisions [13]. A relationship The data provided by this web service are: temperature,
between weather conditions and traffic flow is addressed date and time of data update (refresh occurs
approximately every 30 minutes), a code of the current
in [14], showing a relationship between weather
conditions and traffic speed, as well as a link between climate condition, a broad description of the current
these conditions and the number of accidents. This set climate, a reduced description of the current climate,
generates a change in traffic flow, shown in Fig. 1. current weather information for 'day' or 'n ight', city name
with the code entered, air humidity, wind speed and
sunrise and sunset times, and the general forecast for the
next few days.

3.1.2 Data Collection from the Bus Line

Bus timeliness data on different days and times should be
collected for fo recasting. This collection should be
periodic and occur long enough so that results can be
observed. The authors suggest two years of data co llected
Fig. 1: The relationship between weather, road safety, daily and every 3 minutes, ensuring that a different data
traffic speed and traffic flow. will be acquired fro m the last one, so that the database is
Also in [14], it is shown that on rainy days the number of sufficient fo r use and extract ion of useful data. It is
passengers in buses decreases and the number of cars on suggested that one year of data could already bring
the streets increases. The authors point out that not only satisfactory results. This is possible since the Curitiba
the number of passengers is influenced by the weather but City Hall provides documents and government
also the time it takes to complete the route to its informat ion for web services through an action called
destination and the time waiting for the public transport Open Data Portal [18]. This data is available in open
vehicle. In conclusion, precipitation, cloudiness, wind format for use and unrestricted editing of its users, thus
speed, high temperatures and hail can alter the intensity of being in the public domain and free use, and are intended
traffic and underline the need to incorporate to produce new information and d igital applications for
meteorological conditions into research directly o r society.
indirectly linked to traffic.
The service is in its first version, and it provides databases
Climatic conditions, along with brightness and visibility, of the various organs of the Municipal Govern ment of
and their link to traffic flow and the number of accidents Curitiba. These bases are available through the web site to
are shown in studies in Orange County, Califo rnia [15]. download, or via web services with direct access. The
In this work, some data about the possible influence of informat ion availab le for down load is updated every
the weather conditions are shown in tables, which are month, and can be accessed without the need for a term
checked on some links between the traffic flo w speed and signature or personal identification, with or without
weather conditions. commercial purpose. The information co ming fro m the
web service is released through the delivery of a
document containing the user's login and password by

www.ijaers.com Page | 478

International Journal of Advanced Engineering Research and Science (IJAERS) [Vol -6, Issue-6, June- 2019]
https://dx.doi.org/10.22161/ijaers.6.6.49 ISSN: 2349-6495(P) | 2456-1908(O)
URBS S/A (Co mpany of Urbanization of Cu rit iba), and Web Scraping can be used on real-time map and traffic
has static data, wh ich do not require frequent updating, or sites, since there are many of them that show the current
dynamic, which are updated every 2 minutes, depending flow of traffic in a particu lar location, wh ich could be
on the type of service. For examp le, data fro m the city's used at the time of co llect ion. An examp le is the Google
main po ints of interest are static and do not require Maps tool [22] that reports the time between two points in
updating, since location data and information on the delay real time and whether the traffic flow is flo wing slo wly o r
situation or not on a particular line change at all t imes, quickly. For use in this work it was decided to use only
and therefore have frequent updates. the current time between the arrival and departure points
of the Inter 2 line, that is, when the time was higher than
To request access to the data of a certain line it is
the average the flow is slow and when the arrival time is
necessary to inform the code of the line, which are 3
lower the flow it is faster and therefore faster the bus ride.
characters and can be found in the service itself. When
entering the code line, the following data are available: The current time data between the starting point of the
prefix of the vehicle, which is the specific code of each line and the endpoint was initially acquired using the
vehicle in the network, the time o f the update, latitude and Requests library which is a Python HTTP library that
longitude data in floating point, the line prefix, which is aims to make HTTP requests simp ler and more hu man
the code entered when requesting data, information if are friendly according to documentation in [23]. One of the
adapted for wheelchair users (1 for yes, 0 fo r no), type of uses of the library is the return of the HTML code of the
the bus, the timetable that the vehicle is performing chosen web page and within that code the information of
(normal or Sundays and holidays), a situation of the the time is in the fo rm of text and that piece o f text that is
vehicle timetable (late, early, on time) and the counter of the number of minutes between the points is extracted.
cycles without updating vehicle information, since the This value can then be used to describe the flow o f
informat ion is updated every two minutes. At each cycle current traffic in the region.
of two minutes without update this counter is increased
3.1.4 Additional Data Collection
by 1 (updated information has code 1). The line chosen
for the work was 022 - Inter 2. A data of great importance to the network is the day of
the week in wh ich the collection was performed. The day
3.1.3 Traffic Data Collection
of the week is important because on the Friday before a
Congestion in the city makes everyone involved slow holiday, for examp le, there is a very d ifferent flow o f
down and increase the time spent in traffic. Considering traffic fro m co mmon Tuesdays. First, the day-of-the-week
this, traffic data and traffic flow in the region trafficked data can be obtained in Python (a programming language
by the bus are also important and should be considered. chosen for being one of the options for using Keras that
will be used for the neural network and for having support
One way to collect this data is the Bing Transit [19] web
for all the services used, making only one needed) using
service that also responds with a JSON file with some
the Calendar library. To get the day of the week we
informat ion about accidents or impediments in a
should move the date to a function called "weekday" and
rectangular area fo rmed by two latitudes and two
it returns the day of the week fro m the informed date.
longitudes that represent the four sides of that area. The
Information about special dates or holidays were obtained
following syntax is used to specify this area: a south
using a web service called "Rest-API with Holidays fro m
latitude, a west longitude, a north latitude, and an east
all cit ies of Brazil" [24] and in it is informed the IBGE
longitude. The information provided after specifying the
(Brazilian Institute of Geography and Statistics) code of
region is: t ime and type of accident or impediment (closed
the chosen city and has as return the national, state and
street, construction on the road, collisions of vehicles,
municipal holidays of the city in question. The code of
fallen tree). For use in this work only the nu mber o f
Curitiba is 4106902. When collecting the data a simp le
events in the area was used.
comparison of the current day with the holidays is done to
The Web Scrap ing that was used in this research is a way verify three situations and the answer is transformed into
of requesting data, collecting and analyzing it to extract 3 b its: the first bit fo r holiday eve, another for holiday
desired information by writing a simp le code to perform day, and the last for a day after the holiday, being bit 1 fo r
the task [20]. In [21] it is said that web services are the true answer and bit 0 for false.
standard, in fact, for data collect ion. However, there are
Other data could also influence, such as the occurrence of
scenarios where data is not available through web
large events in the reg ion, and even others of the human
services and the use of Web Scraping becomes necessary.
conviviality itself. Event information could be co llected
in d igital newspapers in the region or on news websites,

www.ijaers.com Page | 479

International Journal of Advanced Engineering Research and Science (IJAERS) [Vol -6, Issue-6, June- 2019]
https://dx.doi.org/10.22161/ijaers.6.6.49 ISSN: 2349-6495(P) | 2456-1908(O)
also through Web Scraping or RSS, but have not been
Condition Code 32
used so far.

3.2 Pre-processing of data Incidents 0

Some data such as the day of the week, the climate Holiday eve 0
description and the bus situation, are in text format and
Holiday 0
should be changed to number, since the neural network
model can use numbers as input data to become more After Holiday 0
optimized. In the first case the following transformation
was made, Sunday for nu mber 1, Monday for nu mber 2, Time 29
Tuesday for number 3 and so on. For the climate
Table 1: A collection of input data held on August 24 at
description the following criterion was adopted, all
noon and eleven minutes.
possible answers were listed and for each one assigned a
number, for examp le "Cloudy weather" was transformed The output data is three, the first one being a bit
into 1 and "Sunny" in nu mber 4. Fo r the situation of the representing delay or not, the second is ahead or not, and
bus the same technique was used, however using 3 the last is on time or not. In no case two of these bits can
numbers each being 0 or 1 depending on the situation, have the value 1, since the bus can not be delayed and
delayed became 100, early 010 and on schedule became advanced at the same time, for example. In Table 2 we
001. see an examp le of output data, where the condition is 100,
that is, delay at the time of collection.
An examp le o f collection is shown in Table 1 and in it the
following data are present: day of the week, which in the Output Data Value
example is the number 6 which is equivalent to a Friday,
day and month of collection which in this case is a day 24 Late 1
of August, the hour and minute of collection, in case
12:11, the temperature in the city at the time of collection, Early 0
29 degrees celsius in the example, also the description of
On Time 0
the current climate, in case 4 that is "sunny", the humidity
of the air, in the collection equal to 40%, condition slug 1 Table 2: A collection of output data held on August 24 at
that is equal to "clear day", then a code of the climate noon and eleven minutes.
condition in question (code generated by the web service When performing the first tests it was observed that it
itself), the nu mber of events collected in the region, would be better to change the qualitative data also for the
holiday and, lastly, the current time between the start and binary form, since the neural network works with weights
end of the line, at that moment was 29 minutes. and sizes when it co mes to numbers. The quantitative data
were kept in their decima l form. Leaving in the
qualitative form might seem to the neural network that
Data Value Monday is less than Saturday for examp le, or that
description 4 is larger than description 1, which is not a
Weekday 6
truth, the idea that should be passed to network is another,
Day 24 it should be something like "it's Monday", yes or no. Then
the following change was made, changing the day -of-
Month 8 week fields, description, quick description and condition
code to a binary form that would be, yes or no for each
Hour 12
possible case, 1 or 0, respectively. For the day of the
Minute 11 week, for examp le, the nu mber of the day has become 7
values, each one being equivalent to one day of the week.
Temp 29 Monday, for examp le, was 1000000, and Tuesday was
0100000. This formatting was used for all cases cited.
Description 4
A data acquisition was done for 3 months, only to verify
Humidity 40 the operation and then continue the data collection,
resulting in that time in 3000 data obtained. After this
Condition Slug 1
collection the data were used to create the neural network.

www.ijaers.com Page | 480

International Journal of Advanced Engineering Research and Science (IJAERS) [Vol -6, Issue-6, June- 2019]
https://dx.doi.org/10.22161/ijaers.6.6.49 ISSN: 2349-6495(P) | 2456-1908(O)
IV.EXPERIMENTS EXECUTION only two layers of training being the first with the input
data and the second with the output with 15 and 3 neurons
In order to imp lement the object ive of predicting delays,
each.
it is necessary to predict events. Prediction is to make
affirmat ions about something that will happen, usually The final configurations used were: two layers (at first the
based on information fro m the past and current state. settings were "normal" in the kernel init ializer option,
Neural networks can be used for predict ion, having "relu" in the activation option, the second the same option
advantages such as automatic learn ing of dependencies, in the kernel init ializer and "softmax" in the activation
requiring only measured data without any need to add option). To co mp ile the model the settings were:
more informat ion. Moreover, the network can be trained "categorical crossentropy" for the loss configuration and
fro m historical data, not having to be represented by an "sgd" optimizer. In the model train ing settings were: 10
explicitly given model. epochs, 100 batch_size and 0 verbose.
4.1 Neural Network

According to [25], Neural Net works, or Artificial Neural 4.3 Results obtained
Networks, find applicat ions in very diverse fields. By
The network, after the train ing, obtained a response with
virtue of their ability to learn fro m input data, with or
84.59% accuracy in the validation data and 92% in the
without a teacher, and by representing a technology
training data so far, that is, the network used some of the
rooted in various disciplines (such as neuroscience, math,
data collected to train and the rest to verify, where 90%
statistics, physics, computer science, and engineering).
was for training. When comparing the results obtained
Some examples o f these fields are modeling, time series
with the results collected in 84.59% of the cases, the
analysis, pattern recognition, signal processing, and
network obtained a correct answer (the highest percentage
control. As stated in [26], artificial neural networks can be
was the correct answer).
considered as a methodology to solve problems
characteristic of artificial intelligence. These data are presented as a chance to occur, for
example, a forecast for June 20 with rain at 12:00 was
Neural networks are massive and parallel systems, made and the following results were obtained: 23.74%
composed of simple processing units that compute certain chances that the bus is late, 9.81% chance of can be early
mathematical functions [27]. Using a set of examples and 66.43% chances that the bus will be on time. So the
presented, the networks are able to generalize the final response fro m the netwo rk is that the bus will
assimilated knowledge to a set of unknown data. They probably be on time on June 20 at noon. The result that
also have the ability to ext ract non-explicit characteristics can be verified in the day and time in question, if the
fro m a set of information provided to them as examples climatic conditions are predicted correctly.
[28].
In view of the results presented here, the network presents
4.2 Experiment setup
a reasonable response considering some field tests with
Keras is described in its documentation [29] as an open positive results and possibly when performing a larger
source neural network library written in Python. It is able data collection the network may present an even better
to work with tools like Google TensorFlow [30]. response.
Designed to enable rapid experimentation of deep neural
V.CONCLUSION
networks, it focuses on being easy to use, modular and
The population satisfaction with public services is fro m
extensible. It is an open source library for nu merical
great importance for imp roving the quality of life,
computation and machine learn ing [31], and used as the
facilitating day-to-day living, and raising the level o f
neural network of this work.
satisfaction with the government. The area of public
To make use of the tools a code in Python language was transportation has a huge problem with delays and
developed, with data input and output in a Comma - requires methods that obtain good accuracy in their
separated values (CSV) file that allo ws the creation of predictions. Considering this need, this work proposed an
tables with data separated by commas. The number o f approach for the city of Curitiba, focused on the
training times was defined, a hit and error quantity collection of informat ion that may be direct ly related to
classifier was created and an interface showing the delays. The proposed approach is based on the collection
response of the system to an input (late, early or on time). of data fro m various mo ments and sources in a way that it
The best result was obtained without changing the makes possible the use of neural networks for prediction.
optimizer and with 10 training periods. The model uses

www.ijaers.com Page | 481

International Journal of Advanced Engineering Research and Science (IJAERS) [Vol -6, Issue-6, June- 2019]
https://dx.doi.org/10.22161/ijaers.6.6.49 ISSN: 2349-6495(P) | 2456-1908(O)
The results achieved have been satisfactory at first and Cambridge Journal of Regions, Economy and Society,
fro m them a more in-depth research can be done, and then 1(3), 477-501.
the data can be distributed to users as a way to improve [10] M oraes Filho, H. D. (2017). Aplicação para
the level of satisfaction with public transportation. monitoramento de frota de ônibus de transportecoletivo no
Distrito Federal.
The paper presented here exposed a methodology for
[11] Serafim, M ., Oliveira, M ., Esteves, D., & Duarte-Neto, P.
collecting data linked to possible changes in the flow of (2014). M odelo de previsão de partida de
buses in the city of Curit iba and how these data can be ônibusutilizandocadeias de M arkov de alcancevariável (pp.
used to predict these variations in advance using neural 624-628). Revista da Estatística da Universidade Federal
networks, notify users and whom they care about. The de Ouro Preto, 3(3).
next objective is to develop a platform in which users [12] Lucio, D. R. (2011). Um aplicativo para
would be able to identify the bus that they will use and dispositivosmóveisvoltado para usuários de
what time they want to arrive and the p latform could transportepúblico (Bachelor's thesis,
notify the user of the ideal time to board the bus or show UniversidadeTecnológica Federal do Paraná).
the user a table with the schedules for departure and [13] Khattak, A. J., & De Palma, A. (1997). The impact of
arrival of the chosen bus. adverse weather conditions on the propensity to change
travel decisions: a survey of Brussels commuters.
ACKNOWLEDGEMENTS Transportation Research Part A: Policy and Practice, 31(3),
We thank the Professors, the Positivo University and the 181-203.
research groups they have created to provide a constant [14] Cools, M ., M oons, E., & Wets, G. (2010). Assessing the
search for knowledge, as well as promot ing and impact of weather on traffic intensity. Weather, Climate,
and Society, 2(1), 60-68.
supporting the growth and professional development of
their students. [15] Golob, T. F., & Recker, W. W. (2003). Relationships
among urban freeway accidents, traffic flow, weather, and
REFERENCES lighting conditions. Journal of transportation engineering,
129(4), 342-353.
[1] Rodrigues, M . A. e Sorratini, J. A. (2008). A qualidade no
transportecoletivourbano. Panorama nacional da [16] M eteorological Data HG Weather - Weather forecast API
pesquisaemtransportes, 1081-1092. [HG Brasil 2018 | HG Weather 2009 – 2018 [Online].
Retrieved from https://hgbrasil.com/
[2] Sá, D., & Rezende, D. A. (2019). Economia Da Inovação
No Contexto Da Cidade Digital Estratégica: Caso Do [17] Tardy, C., M occozet, L., &Falquet, G. (2016, April). A
M unicípio De Curitiba. simple tags categorization framework using spatial
PerspectivasemGestão&Conhecimento, 9(1), 62-81. coverage to discover geospatial semantics. In Proceedings
of the 25th International Conference Companion on World
[3] M aciel, M . D. A. (2016). Previsão de horários dos ônibus
Wide Web (pp. 657-660). International World Wide Web
do sistema de transportepúblicocoletivo de Campina
Conferences Steering Committee.
Grande.
[18] Open Data Portal - Curitiba City Hall. [Online]. Retrieved
[4] Gama, K., Alvaro, A., & Peixoto, E. (2012). Emdireçãoa
from www.curitiba.pr.gov.br/dadosabertos/
um modelo de maturidadetecnológica para
cidadesinteligentes. SimpósioBrasileiro de Sistemas de [19] M icrosoft, Inc. Bing M aps APIs. [Online]. Retrieved from
Informação, VIII. https://msdn.microsoft.com/pt-br/library/dd877180.aspx
[5] Fonseca, R., M arques, M . C. &Velloso E. (2015). Retratos [20] M itchell, R. (2018). Web Scraping with Python: Collecting
da SociedadeBrasileira - M obilidadeurbana. ISSN 2317- M ore Data from the M odern Web. " O'Reilly M edia, Inc.".
7012. Ano 5. Número 27. [21] Glez-Peña, D., Lourenço, A., López-Fernández, H.,
[6] Resende, P. D. T. V., & Sousa, P. R. D. (2009). Reboiro-Jato, M ., &Fdez-Riverola, F. (2013). Web
M obilidadeurbananas grandescidadesbrasileiras: um scraping technologies in an API world. Briefings in
estudosobreosimpactos do congestionamento. SIM POI– bioinformatics, 15(5), 788-797.
SIM PÓSIO DE ADMINISTRAÇÃO DA PRODUÇÃO, [22] Google M aps - Navigation and public transport. [Online].
LOGÍSTICA E OPERAÇÕES INTERNACIONAIS, FGV. Retrieved from www.google.com.br/maps/
[7] Bertini, R. L. (2006). You are the traffic jam: an [23] Reitz K. Requests, Python HTTP for Humans. (2018).
examination of congestion measures. Transportation [Online]. Retrieved from http://docs.python-
Research Board, Washington DC. requests.org/en/master/user/quickstart/
[8] Rufolo, A. M ., & Bianco, M . J. (1998). The impact of [24] API Rest-full web service - Holidays from all Brazilian
congestion pricing and parking taxes on spatial cities. [Online]. Retrieved from
competition. http://www.calendario.com.br/
[9] Van der Ploeg, F., &Poelhekke, S. (2008). Globalization [25] Haykin, S. (2007). Redes neurais: princípios e prática.
and the rise of mega-cities in the developing world. Bookman Editora.

www.ijaers.com Page | 482

International Journal of Advanced Engineering Research and Science (IJAERS) [Vol -6, Issue-6, June- 2019]
https://dx.doi.org/10.22161/ijaers.6.6.49 ISSN: 2349-6495(P) | 2456-1908(O)
[26] Barreto, J. M . (2002). Introduçãoàs redes neuraisartificiais.
V Escola Regional de Informática. SociedadeBrasileira de
Computação, Regional Sul, Santa M aria, Florianópolis,
M aringá, 5-10.
[27] Braga, A. D. P., Carvalho, A. P. L. F. &Ludemir, T. B.
(1998). Fundamentos de redes neuraisartificiais. Rio de
Janeiro: 11a Escola de Computação.
[28] Kovács, Z. L. (2002). Redes neuraisartificiais.
EditoraLivraria da Física.
[29] Keras: The Python Deep Learning library. [Online].
Retrieved from https://keras.io/
[30] Google TensorFlow [Online]. Retrieved from
https://www.tensorflow.org/
[31] Araújo, F. H., Carneiro, A. C., Silva, R. R., M EDEIROS,
F. N., & USHIZIM A, D. M . (2017). Redes
NeuraisConvolucionais com Tensorflow: Teoria e Prática.
SOCIEDADE BRASILEIRA DE COM PUTAÇÃO. III
Escola Regional de Informática do Piauí. Livro Anais-
Artigos e M inicursos, 1, 382-406.

www.ijaers.com Page | 483