=Paper=
{{Paper
|id=Vol-2492/paper8
|storemode=property
|title=A Machine Learning Approach to Monitor Air Quality from Traffic and Weather Data
|pdfUrl=https://ceur-ws.org/Vol-2492/paper8.pdf
|volume=Vol-2492
|authors=Claudio Rossi,Alessandro Farasin,Giacomo Falcone,Carlotta Castelluccio
|dblpUrl=https://dblp.org/rec/conf/ami/RossiFFC19a
}}
==A Machine Learning Approach to Monitor Air Quality from Traffic and Weather Data==
https://ceur-ws.org/Vol-2492/paper8.pdf
A Machine Learning Approach to Monitor Air Quality
from Traffic and Weather data
Claudio Rossi Alessandro Farasin
LINKS Foundation, Italy Politecnico di Torino, Italy
claudio.rossi@linksfoundation.com & LINKS Foundation, Italy
alessandro.farasin@polito.it
Giacomo Falcone Carlotta Castelluccio
LINKS Foundation, Italy Microsoft, Italy
giacomo.falcone@linksfoundation.com carlotta.castelluccio@microsoft.com
Abstract.
Knowing the amount of air pollutants in our cities is of great impor-
tance to help decision makers in the definition of effective strategies
aimed at maintaining a good air quality, which is a key factor for a
healthy life, especially in urban environments. Using a data set from a
big metropolitan city, we realize the uAQE: urban Air Quality Eval-
uator, which is a supervised machine learning model able to estimate
air pollutants values using only weather and traffic data. We evaluate
the performance of our solution by comparing the predicted pollutant
values with the real measurements provided by professional air moni-
toring stations. We use the predicted pollutants to compute a standard
Air Quality Index (AQI) and we map it into a set of five qualitative
AQI classes, which can be used for decision making at the city level.
uAQE is able to predict the AQI class value with an accuracy of 0.8.
Keywords: Air Quality · Machine Learning · BRNN · Weather · Traffic
1 Introduction and Related Works
Air pollution introduces into the atmosphere chemicals, particulates, or biological materials that causes discom-
fort, disease, or death to humans and to other living organisms alike. More than 5.5 million people worldwide
are dying prematurely every year as a result of air pollution exposure [1]. This fact confirms that air pollution
is one of the world’s largest environmental health risks. Most of these deaths are occurring in rapidly developing
economies, e.g., China and India, but also in European metropolitan cities, e.g., Naples, Turin and Milan, which
have an air pollution index among the highest ones according to recent rankings 1 .
Road transport is one of the main causes of air pollutants emissions, accounting for the 14% of the total emis-
sions in European countries 2 . In recent years, advanced after-treatment technology (Particulate Matter traps)
Copyright © by the paper’s authors. Use permitted under Creative Commons License Attribution 4.0 International (CC
BY 4.0).
In: I. Chatzigiannakis, P. Ciampolini, S. Hanke, G. Mylonas: Proceedings of the BRAINS Workshop, Rome, Italy, 13-
Nov-2019, published at http://ceur-ws.org
1
http://www.numbeo.com/pollution/rankings.jsp
2
http://www.eea.europa.eu
1
have been implemented by car manufacturers in response to increasingly tight European emission policies. Con-
sequently, emissions of the main regulated pollutants from road transport, i.e., Nitrogen Oxides (N Ox ) and
Particulate Matter (PM), have been reduced despite the increased vehicle activity. Specifically, urban N Ox
emissions from traffic has been reduced by 16% between 2000 and 2010, mainly due to the introduction of the
Euro 4 and the Euro 5 standards for passenger cars (both petrol and diesel), which were applied from early 2005
and late 2009, respectively [3].
Other human activities having a strong impact on air quality are industrial processes, farming, heat and air
conditioning, and other types of transport (trains, airplanes,etc.).
It is a well-known fact that weather phenomena have a strong impact on air pollutants because once pollutants
are emitted into the air, they propagate into the atmosphere according to weather conditions, e.g., turbulence
mixes pollutants into the surrounding air, and wind carries them away from the source location. Conversely, when
the air near the surface of the earth is cooler than the air above (a phenomenon called temperature inversion)
there is very little air mixing. Since cool air is heavy, it will not to move up to mix with the warmer air above.
Thus, any pollutants released near the surface will get trapped and build up in the cooler air layer.
Municipalities struggle to predict the effect of traffic policies, e.g., total traffic block, stop of most pollutant vehi-
cles, on the air quality because there is a lack of easy-to-use tools that can estimate the air pollution taking into
account also the meteorological predictions. Furthermore, the availability of air quality measurement stations
in a city is very limited due to economic constrains. A professional station requires a non negligible investment
(about 200k e per installation) and it has a high maintenance cost (about 30ke per year) [2].
Because of its importance, the estimation of the air quality has been subject to some studies. In [2], Microsoft
researchers proposed a semi supervised learning approach able to predict P M10 and Nitrogen Dioxide (N O2 )
emissions at an higher spatial resolution with respect to the one achieved by the installed air quality sensors by
coupling other data sources such as traffic flows, the structure of the road network, meteorological conditions
and point of interest locations. Their solution is complementary to ours, and it can can be used to improve the
spatial resolution of the uAQE. Other relevant studies include the [4] and [5], which present a set of learning
methods able to predict N Ox concentrations from past observations and weather conditions. In [6], the authors
studied Delhi’s P M2.5 concentrations and its correlation with the vehicular traffic and with the weather condi-
tions. However, the proposed model makes several empirical assumptions and it includes parameters specific to
the city of Delhi. Hence, it cannot be re-used for our purpose.
To help decision maker in keeping under control the air quality we propose uAQE: urban Air Quality Evaluator,
which is a set of supervised machine learning model able to predict air pollutants values in a urban environment
using only weather and traffic data. We train our models with data taken from Milan, building one model for
each air pollutants. Our work is different from all the above mentioned approaches because we aim to predict
pollutants without requiring data from air quality stations. Note that we train one model for each air pollu-
tants, namely Nitrogen Dioxide (N O2 ), Ozone (O3 ), Carbon Monoxide (CO), Benzene (C6 H6 ), Total Nitrogen
(N2 ), Particulate Matter (P M10 ), Sulfur Dioxide (SO2 ), Particulate Matter (P M2.5 ), Black Carbon (BC), and
Ammonia (N H3 ). We present the accuracy of each model using the pollutants as measured by professional air
stations. Following a regional standard, we use the predicted pollutants to compute an Air Quality Index (AQI)
which is then mapped it into a set of five qualitative classes that are used to manage air quality policies at city
level. We finally asses the classification accuracy achieved by uAQE obtaining a value of 0.8.
2 Input data
Our data has been collected in the city of Milan during two months (Nov.- Dec. 2013), and it contains three
distinct data categories:
– Weather: we have six different weather stations placed within the city limit. Each station has a unique ID,
type, location, and it features a set of co-located sensors. Each sensor measures a different meteorological
phenomena. This information has been obtained thanks to ARPA (Agenzia Regionale per la Protezione
dell'Ambiente) 3 .
– Traffic: through fixed video cameras already installed for traffic access control at 52 locations in the central
area of Milan (Cerchia dei Bastioni) the local authority obtained the plate number of transiting vehicles,
from which the vehicle characteristic could be extracted from the official database, i.e., the Motorizzazione
civile, which holds the information of all Italian vehicles. Note that we received anonymized data, i.e., with
3
http://ita.arpalombardia.it/ITA/qaria/doc RichiestaDati.asp
2
Fig. 1. Sensors locations map. White squares are fixed cameras (traffic gates), meteo icons represent weather station, and
black icons are air stations.
hashed plate numbers and with no information about the vehicle owner. Therefore, only the technical details
of each vehicle has been made available to us. These data have been provided as open data by the city of
Milan.
– Air: we take the measurements of three different air stations located within the city limits. Each station
features multiple co-located sensors, each of which measures a single air pollutant. Also these measurements
are directly provided as open data by ARPA, who is the official source of this kind of data.
The locations of weather stations, air stations and fixed cameras are shown in Fig. 1.
The weather station data contain wind direction (degree), wind speed (m/s), temperature (Celsius degree),
relative humidity (%), precipitation (mm), global radiation (µW/m2 ), net radiation (µW/m2 ), and atmospheric
pressure (hPa). The traffic data include each vehicle passage at each gate, for which the location and the
timestamp of each passage is known. For each passage, the vehicle characteristics are given, namely the European
emission standard category (EURO category from 1 to 6), the vehicle type (i.e., bus, freight, transport, people
transport or not available), the fuel type (i.e., petrol, diesel, electric, LPG, hybrid or missing), the presence of
the Diesel Particle Filter (DPF) and the vehicle length expressed in mm.
The air pollution data contain ten different agents: N O2 (µg/m3 ), N H3 (µg/m3 ), C6 H6 (µg/m3 ), SO2
(µg/m3 ), BC (µg/m3 ), CO (µg/m3 ), N2 (ppb), P M10 (µg/m3 ), P M2.5 (µg/m3 ), O3 (µg/m3 ).
We compute the hourly air quality index as defined by Piedmont index AQI because is the only example of
operational use of an air quality index in Italy4 . The AQI uses only three pollutants, namely N O2 , P M10 , O3 ,
and it is formulated as follows:
Vmed24hP M10
IP M10 = × 100 (1)
VrifP M10
VmaxhN O2
IN O2 = × 100 (2)
VrifN O2
Vmax8hO3
I8hO3 = × 100 (3)
Vrif8hO
3
IP M10 + max(IN O2 , IO3 )
IAQI = (4)
2
We observe that in the considered data set O3 never exceeds the maximum value established for preserving
human health (i.e., 120µg/m3 ), whereas N O2 exceeds its hourly maximum value (i.e., 200µg/m3 ) only in few
cases (< 5%). Conversely, P M10 exceeds the the daily maximum value (i.e., 50µg/m3 ) in 50% in the cases.
We map the computed AQI in the five classes defined by the Piedmont region, namely Optimal (0 ≤ AQI < 50),
Good (50 ≤ AQI < 75), Fair (75 ≤ AQI < 100), Average (100 ≤ AQI < 125), Not Very Healthy (125 ≤ AQI <
4
http://www.arpae.it/cms3/documenti/aria/IQA.pdf
3
Fig. 2. Temporal evolution of the Air Quality Index computed in the considered time frame.
150), Unhealthy (150 ≤ AQI < 175), Very Unhealthy (AQI ≥ 175). In our data set, we observe that there are
no AQI values in the Optimal level and very few in the Good one, while the most part of values (≈ 80%) fall
between Fair and Not Very Healthy levels. We show in Fig. 2 the temporal evolution of the AQI.
3 Feature Construction
Our aim is to use a supervised machine learning approach to predict pollutants from traffic and weather data
under the hypothesis that we do not have air sensors to directly measure air pollutants. We show the empirical
cumulative distribution function (ecdf) of the considered weather data in Fig.3 and the distribution of the
traffic features in Fig.4. For sake of readability, the ecdfs related to the air pollutants are shown in Fig.7, in the
Appendix.
The aforementioned data categories, namely weather, traffic and air pollutants, are merged at hourly resolution,
according to the maximal temporal resolution of both weather and air data. Then, measurements produced by
different sensors of the same type in the same hour are averaged.
Finally, the hourly passages at vehicle gates are counted, according to four different sets: (i) the EURO class
(EURO), (ii) the vehicle type (Vtype), (iii) the fuel type (Ftype), (iv) existence of Diesel Particulate Filter
(DPF).
In order to perform all data manipulations, we use R and the plyr library, which provides data aggregation
operators. As a final step, we filled a small percentage of missing values (< 1%) in the air and weather data
by polynomial interpolation using the spline function of the zoo library. Conversely, because of the remarkable
percentage of NA values for each group in the traffic features and due to not available information, we filled
missing values following the probability distribution of data.
The final feature set is composed by the following variables:
– Time: day of week (1-7), hour (1-24). This is to consider the regular patterns of human activities, which are
framed within the day and within the week;
– Hourly passages: counts of total passages and aggregated count by EURO class, vehicle type, fuel type,
existence of particulate filter. Because we compute the total passages, we remove one category from each
aggregation to avoid creating features which are linear combination of other ones while reducing the number
of total features;
– Hourly weather phenomena averages: wind direction, wind speed, temperature, relative humidity, pre-
cipitation, atmospheric pressure.
In order to consider the effect of the past on the current pollutants levels, for each traffic and weather feature
f (t) (wind direction excluded) we add another feature f p(t) equal to the sum of f (t) over the last x time slots
Pt=i−x
(f p(t) = t=i f (t)). We evaluated increasing values of x starting from 1 and and we empirically found the
best value to be 12.
Studying the correlation between weather and pollutants (Fig. 5) we noticed that temperature, relative humidity,
precipitation wind speed, and atmospheric pressure are the most significant ones, having an average absolute
4
Fig. 3. Weather features Ecdf Graphs
Fig. 4. Traffic Features Distribution
correlation of 0.40, 0.20, 0.13, 0.51, 0.51 with the pollutants considered in the AQI computation, respectively.
Conversely, all traffic features results less correlated and they are not shown for brevity.
4 Pollutants prediction and and evaluation of AQI
We implemented several machine learning models using the caret package. In particular, we tested algorithms for
regression, including Generalized Linear Model (GLM), Random Forest (RF), Support Vector Machines (SVM)
and Artificial Neural Networks (ANN). We tested all algorithms with default hyper-parameters and with the same
random seed, uniformly selecting in time the 70% of the samples as training set, and leaving the remaining 30%
for the test set. We trained all models with a 5-fold cross-validation and we computed the model performances in
terms of mean squared error (MSE) for pollutants. For brevity, we report only the results of the GLM, which we
considered as the baseline and of the ANN, which is the model that performed best. Artificial Neural Network
(ANN) [8] are inspired by biological nervous systems such as the human brain, which process information through
a large number of highly interconnected processing elements (neurons). ANNs can be used in several applications,
such as pattern recognition or data classification, and they are a supervised machine learning technique.
Specifically, we chose a particular type of ANN, namely the BRNN model (Bayesian Regularization of Neural
5
Fig. 5. Correlation between selected weather features and pollutants used for the AQI computation.
Table 1. Performance comparison between the GLM and the BRNN reporting average and standard deviation of the
absolute error for each pollutant.
BRNN 9 Neur. GLM Comparison
Agent Unit µ(ε) δ(ε) µ(ε) δ(ε) ∆[µ(ε)] ∆[δ(ε)]
NO2 µg/m3 12.45 10.28 21.02 20.58 41% 50%
3
O3 µg/m 22.47 20.06 46.47 45.09 52% 56%
3
CO µg/m 10.99 10.18 19.31 15.59 43% 35%
3
C6H6 µg/m 39.85 64.44 98.92 145.89 60% 56%
N2 ppb 27.33 29.12 56.27 81.32 51% 64%
3
PM10 µg/m 13.65 14.35 34.56 34.62 61% 59%
3
SO2 µg/m 18.64 20.67 38.58 42.68 52% 52%
PM2.5 µg/m3 13.28 13.88 32.55 31.87 59% 56%
3
BC µg/m 18.47 24.21 35.91 66.30 49% 63%
3
NH3 µg/m 26.95 52.95 41.83 119.12 36% 56%
Networks) [7], because it is more robust than standard back-propagation networks and it can reduce the need for
lengthy cross-validation. Bayesian regularization is a mathematical process that converts a nonlinear regression
into a “well-posed” statistical problem in the manner of a ridge regression [7]. The main model parameter is the
number of neurons n to be used. In order to define the optimal n, we incrementally evaluated the model accuracy
starting with n = 1 and incrementing it in steps of 1 until 20. Therefore, we empirically find the best value of
n = 9, after which the performance improvement can be considered negligible.
For each pollutant, we compare the BRNN model with the GLM performances, obtaining for the BRNN an
6
Table 2. Confusion Matrix of the Air Quality Index class computed with the pollutants estimated with the BRNN model.
Real
Predicted Good Fair Average Not Very Healthy Unhealthy Very Unhealthy
Good 72 0 0 0 0 0
Fair 120 528 0 0 0 0
Average 0 69 312 0 0 0
Not Very Healthy 0 0 0 144 0 0
Unhealthy 0 0 0 72 120 0
Very Unhealthy 0 0 0 0 24 0
Overall Statics Accuracy: 0.8049
improvement of the average relative error between 36% and 61% over the GLM. The performance comparison is
fully reported in Table 1.
Using the predicted values of N O2 , O3 and P M10 , we computed the AQI value, and then we mapped it into the
classes described in Section 2 (i.e. Optimal, Good, Fair, Average, Not Very Healthy, Unhealthy, Very Unhealthy).
Finally, we computed the accuracy of the estimated AQI class, which is reported in Table 2.
Fig. 6. Scatter Plot AQI measured/predicted
Our model predicts AQI with a class accuracy of 0.8, which we evaluate as satisfactory (see also the scatter plot
in Fig.6), especially considering that the distance of the classification error is never greater than one, meaning
that when the model predicts an erroneous class it is never beyond the adjacent one, e.g., the model can predict
Fair instead of Good but it never predicts Average or any worst condition instead of Good.
5 Conclusion and Future works
In this paper we used traffic and weather data in order to predict the air pollution in a metropolitan city. We
designed and implemented a set of machine learning models to predict single pollutants that we used to compute
qualitative air quality classes based on a standardized Air Quality Index (AQI). The performance of our best
model, (BRNN with 9 neurons), achieves an AQI class accuracy of 0.8.
Future works will include the evaluation of our approach on a bigger dataset, an improvement of the feature
set, and the evaluation of several scenarios (e.g., including partial or complete traffic block, different weather
conditions, etc.) in order to evaluate the impact of local traffic policies on the air quality.
7
References
1. J. Amos, “Polluted air cause 5.5 million deaths a year new research says”, BBC NEWS, Science and Environment,
2016.
2. Y. Zheng,F. Liu,H. Hsieh,“U-air: When Urban Air Quality Inference Meets Big Data”, Microsoft Research Asia, ACM,
2013
3. I. Sundvor, N. Castell Balaguer, M. Viana, X. Querol, C. Reche, F. Amato, G. Mellios, C. Guerreiro,“Road traffics
contribution to air quality in European cities”, ETC/ACM, 2012
4. I. Juhosa, L. Makrab, B. Ttha, “Forecasting of traffic origin NO and NO2 concentrations by Support Vector Machines
and neural networks using Principal Component Analysis”, Simulation Modelling Practice and Theory, vol. 16, no. 9,
pp. 1488-1502, 2008.
5. R. Berkowicz , F. Palmgren, O. Hertel, E. Vignati, “A Study on Effects of Weather, Vehicular Traffic and Other Sources
of Particulate Air Pollution on the City of Delhi for the Year 2015”, Journal of Environment Pollution and Human
Health, vol. 4, no. 2, pp. 24-41, 2016.
6. R. Gopalaswami,“Using measurements of air pollution in streets for evaluation of urban air quality meterological
analysis and model calculations”, Science of The Total Environment, vol. 189-190, pp. 259-265, 1996.
7. F. Burden,D. Winkler,“Bayesian regularitazion of neural networks”, PubMed, 2008
8. C. Stergiou, D. Siganos ,“Neural networks”, Imperial College London, 2011.
Appendix: Pollution agents Ecdf Graphs
Fig. 7. Pollution Agents Ecdf Graphs
8