# Application of Machine Learning to Include Honking Effect in Vehicular Traffic Noise Prediction

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

## Featured Application

**Machine learning techniques are calibrated on a dataset of sound levels in urban areas in India, with relevant honking occurrences. The best model chosen will improve the prediction of road traffic noise in similar cases, with respect to standard models that neglect the effects of honking.**

## Abstract

_{eq}A method to include the honking effect in the traffic noise prediction has been illustrated. The techniques that have been used for the prediction of traffic noise are decision trees, random forests, generalized linear models and artificial neural networks. The results obtained by using these methods have been compared on the basis of mean square error, correlation coefficient, coefficient of determination and accuracy. It has been observed that honking is an important parameter and contributes to the overall traffic noise, especially in congested Indian road traffic conditions. The effects of honking noise on the human health cannot be ignored and it should be included as a parameter in the future traffic noise prediction models.

## 1. Introduction

_{2}) emissions and unnecessary fuel consumption are addressed in the framework of route management for autonomous vehicles in urban areas. Conversely, there is a relatively lesser awareness of the harmful effects of traffic noise on the human population. Some of the adverse effects [5,6,7,8] include sleep disturbance, speech interference, annoyance, cardio-vascular effects and loss of fertility. In rare cases, such as with traffic policemen who are exposed to high sound pressure levels and long-term road traffic noise, hearing loss has also been observed [9]. As for the prediction of traffic volumes and the adoption of traffic management strategies to reduce noise pollution levels, a lot of research has been done on the effects of noise on human health [10,11]. Traffic noise prediction models play a very important role not only in the design and modification of road and traffic infrastructure [12] but also in the assessment of the noise level, based on certain critical parameters. Different types of traffic noise prediction models include country or region-specific models like the “Federal Highway Administration model” (FHWA, USA), the “UK Calculation of Road Traffic Noise” (CoRTN, UK), the “Acoustical Society of Japan Road Traffic Noise model” (ASJ, Japan) and the “Richtlinien für den Lärmschutz an Straben” (RLS-90, Germany). A critical review of the most used models can be found in [13,14,15,16]. Other categories of models include static and dynamic, stochastic and deterministic and artificial intelligence or machine learning based models [17,18,19,20,21,22]. In particular, several applications to traffic noise prediction can be found in literature, for instance [23,24,25,26,27], in which the authors adopt artificial intelligence techniques, such as artificial neural networks (ANN) and genetic algorithms (GA).

_{den}) up to 0.5–13 dBA compared to homogenous traffic conditions [31]. Therefore, the prediction models used for homogenous traffic conditions are not applicable in this heterogeneous context. Thus, to increase the accuracy of noise prediction models, considering honking is required.

## 2. Materials and Methods

#### 2.1. Dataset and Sites Description

_{eq}), the data collection included the parameters traffic volume (Q), percentage of heavy vehicles (P) and total honking occurrences (H) that have been used to calibrate and test the machine learning models. These parameters have been collected manually using videography at five different identified sites, highlighted in Figure 1, on the basis of congestion, presence of honking noise and proximity to resident population. For each video, the numbers of light and heavy vehicles have been annotated, as well as the number of honks in each 15 min measurement.

_{eq}in dBA) have been experimentally measured using a sound level meter (SLM, B&K make, 2250). The sound level meter used was a class 1 integrating type, which meets with the IEC specifications (IEC 61672-1: 2002, International Electrotechnical Commission, 2002) [33]. The SLM was mounted on a tripod (in order to avoid the human body impedance effects), at a height of 1.2 m above the ground level [34] and at a distance of 1 m from the edge of the road where the traffic noise measurements were taken.

#### 2.2. Bivariate Correlation Analysis

_{eq}) are reported with bivariate scatter plots below the diagonal, histograms on the diagonal and the Pearson correlation above the diagonal. A high correlation (0.90) is found between L

_{eq}and traffic volume, as expected since the main noise sources in the measurement sites are the vehicles. A significant (0.65) correlation is found between L

_{eq}and P, as well as between L

_{eq}and H.

#### 2.3. Machine Learning Methodologies

- (i).
- Decision Trees (DT) [18]: These methodologies are employed in machine learning applications where an analysis of the data is required. They use a structure which resembles a flow chart as they are based on deterministic data structures and used in classification problems. At the top of the tree, there is a root node and the branches represent the tests that are done and the leaves denote the results of the tests. The rpart( ) function builds a decision tree model. A decision tree works by splitting nodes into sub-nodes. The parameter MinSplit describes the minimum number of members that a node should have before the split is attempted. MaxDepth indicates the maximum depth or length of the tree, starting from the root node up to the leaf node. MinBucket parameter specifies the minimum number of entities that a leaf node can have. The default value is generally one-thirds of the MinSplit value.
- (ii).
- Random Forests (RF) [36]: In this approach, groups of decision trees are created. This is an ensemble method, which can be used for regression, classification and other tasks. They avoid over-fitting, which can be a drawback in the decision tree method, by making random decision forests. The observations are used as input for each tree and the most common outcome is used as the final output. As the errors are cancelled out, a more accurate prediction is obtained using this method. The randomForest( ) function is used for the implementation of the random forests. Breiman (2001) [36] introduced the idea of random sampling of variables at each node as the tree is being built. He also introduced the bagging concept for the sampling [38] in which random samples are chosen for the training dataset for each tree. It helps in making the model robust to noise and outliers. The parameter ntree specifies the number of trees built in the Random Forests model.
- (iii).
- Linear Models: Generalized linear models (GLM) make use of and combine different types of regression, e.g., linear and logarithmic [39]. They take care of different types of distribution like the log-linear and log-odds, as the response is not always linear and might not follow a normal distribution.
- (iv).
- Neural Networks [40]: Artificial neural networks (ANN) are similar in working to the human brain which uses neurons (connections of nodes) for the learning tasks. The process involves the assignment of weights to the inputs and activation functions for getting the desired outputs. There are different layers in a neural network like an input layer, a hidden layer (which performs non-linear transformations of the inputs) and an output layer of neurons. The function neuralnet( ) is used to build a neural network model in R software [35]. The parameter hlayers is used to specify the number of hidden layer nodes or neurons in the NN architecture. MaxNWts sets the maximum limit of the number of weights that can be used in the model. The maxit parameter defines the maximum number of iterations to be done during the training.

#### 2.4. Perfomance Metrics

^{2}), mean square error (MSE) and accuracy, briefly resumed in this subsection.

#### 2.4.1. Correlation Coefficient (r)

#### 2.4.2. Coefficient of Determination (R^{2})

^{2}provides the percentage variation in Y explained by X-variable. It is the square of the coefficient of correlation (r) therefore it is a measurement used to explain how much variability of one factor can be caused by its relationship to another related factor. When the coefficient of determination is equal to 1, the regression line fits all the sample data [42].

#### 2.4.3. Mean Squared Error (MSE)

#### 2.4.4. Accuracy

## 3. Results and Model Comparison

^{2}, MSE and accuracy. The above relation has been used for an accuracy of ±1 dBA. The values of the different criteria for the four models are shown in Table 3.

^{2}, MSE and accuracy. The value of MSE is 0.413 for rf which is the lowest and the value of accuracy is 94% (for ±1 dBA) which is the highest among all other models.

_{eq}, all the models’ results are gathered almost close to the bisector, meaning a general effectiveness of the proposed models. When the measured L

_{eq}are around 70 dBA (site 2), the models have the best performances. This site is characterized by absence of heavy vehicles and honking frequencies lower than in the other sites. Data above 78 dBA are related to site 4, in which the traffic flows are higher than in the other sites.

_{eq}for the four methods is shown in Figure 5.

## 4. Discussion

_{eq}and all the models’ results are gathered almost close to the bisector, meaning a general effectiveness of the proposed models. When the measured values of L

_{eq}are around 70 dBA (site 2), the models have the best performance. This site is characterized by the absence of heavy vehicles and honking occurrences lower than in the other sites.

^{2}basically constant, the accuracy has a little increase from 60.0% to 66.7%, but MSE increases from 0.797 dBA to 0.809 dBA. The other two models, namely GLM and ANN, converge to similar predictions and, consequently, to equal performance metrics. They exhibit a worsening of all the selected metrics when honking is not considered. Since GLM was the best performing model in the training dataset with honking, it is evident that the inclusion of this parameter leads to a better prediction of noise levels in the presented application. When honking is not included, the MSE of GLM increases from 0.666 dBA to 1.291 dBA and the GLM accuracy gets worse, lowering from 80% to 46.7%.

## 5. Conclusions

^{2}, mean square error and accuracy. It is seen that the generalized linear model (GLM) and artificial neural networks (ANN) are the best performing models, with GLM doing slightly better than ANN on the considered criteria.

## Author Contributions

## Funding

## Data Availability Statement

## Acknowledgments

## Conflicts of Interest

## Appendix A

**Table A1.**Full dataset containing traffic volumes (Q), percentage of heavy vehicles (P), total honking occurrences in 15 min (H) and continuous equivalent sound level (L

_{eq}) for all the measurements, in all the sites.

Progressive Number | Traffic Volume (Q) [veh] | Percentage of Heavy Vehicles (P) [%] | Total Honking Occurrences (H) [Counts] | Continuous Equivalent Sound Level (L_{eq}) [dBA] | |
---|---|---|---|---|---|

Site 1 | 1 | 570 | 3.3 | 110 | 74.0 |

2 | 582 | 2.2 | 135 | 74.6 | |

3 | 574 | 3.7 | 112 | 74.5 | |

4 | 576 | 2.8 | 118 | 74.7 | |

5 | 585 | 1.7 | 267 | 74.4 | |

6 | 589 | 2.5 | 249 | 74.9 | |

7 | 572 | 3.4 | 128 | 75.1 | |

8 | 584 | 1.3 | 175 | 74.5 | |

9 | 593 | 2.4 | 191 | 74.3 | |

10 | 597 | 1.5 | 231 | 75.0 | |

Site 2 | 11 | 216 | 0.0 | 126 | 69.9 |

12 | 241 | 0.0 | 102 | 70.0 | |

13 | 238 | 0.0 | 110 | 69.7 | |

14 | 239 | 0.0 | 102 | 69.5 | |

15 | 220 | 0.0 | 106 | 68.0 | |

16 | 230 | 0.0 | 115 | 69.3 | |

17 | 226 | 0.0 | 100 | 69.5 | |

18 | 244 | 0.0 | 123 | 70.2 | |

19 | 219 | 0.0 | 93 | 68.6 | |

20 | 246 | 0.0 | 90 | 69.8 | |

Site 3 | 21 | 513 | 0.0 | 363 | 76.5 |

22 | 423 | 0.6 | 293 | 73.0 | |

23 | 419 | 0.7 | 316 | 75.0 | |

24 | 505 | 0.4 | 322 | 75.2 | |

25 | 464 | 0.6 | 330 | 76.0 | |

26 | 463 | 0.2 | 278 | 72.2 | |

27 | 450 | 0.5 | 267 | 72.8 | |

28 | 422 | 0.1 | 161 | 72.3 | |

29 | 433 | 0.2 | 210 | 72.5 | |

30 | 369 | 0.8 | 135 | 72.0 | |

Site 4 | 31 | 1092 | 3.8 | 204 | 78.7 |

32 | 1315 | 3.7 | 209 | 79.7 | |

33 | 1143 | 3.4 | 205 | 78.9 | |

34 | 1141 | 3.6 | 210 | 78.5 | |

35 | 1350 | 3.3 | 232 | 79.5 | |

36 | 1169 | 3.5 | 257 | 78.8 | |

37 | 1344 | 3.2 | 229 | 79.1 | |

38 | 1090 | 3.2 | 268 | 78.5 | |

39 | 1279 | 3.7 | 237 | 79.3 | |

40 | 1377 | 3.1 | 279 | 79.7 | |

Site 5 | 41 | 657 | 0.5 | 406 | 78.0 |

42 | 663 | 0.5 | 229 | 75.1 | |

43 | 687 | 0.4 | 351 | 77.2 | |

44 | 686 | 0.5 | 368 | 77.4 | |

45 | 671 | 0.4 | 257 | 75.7 | |

46 | 689 | 0.4 | 273 | 76.1 | |

47 | 688 | 0.4 | 345 | 77.4 | |

48 | 680 | 0.5 | 360 | 77.8 | |

49 | 675 | 0.5 | 299 | 76.6 | |

50 | 690 | 0.4 | 225 | 75.5 |

## References

- Moshammer, H.; Panholzer, J.; Ulbing, L.; Udvarhelyi, E.; Ebenbauer, B.; Peter, S. Acute Effects of Air Pollution and Noise from Road Traffic in a Panel of Young Healthy Adults. Int. J. Environ. Res. Public Health
**2019**, 16, 788. [Google Scholar] [CrossRef][Green Version] - Gieseke, J.; Gerbrandy, G.J. Report on the Inquiry into Emission Measurements in the Automotive Sector; Committee of Inquiry into Emission Measurements in the Automotive Sector, European Parliament: Brussel, Belgium, 2017. [Google Scholar]
- Pascale, A.; Fernandes, P.; Guarnaccia, C.; Coelho, M.C. A study on vehicle Noise Emission Modelling: Correlation with air pollutant emissions, impact of kinematic variables and critical hotspots. Sci. Total Environ.
**2021**, 787, 147647. [Google Scholar] [CrossRef] - Zambrano-Martinez, J.L.; Calafate, C.T.; Soler, D.; Lemus-Zúñiga, L.G.; Cano, J.C.; Manzoni, P.; Gayraud, T. A centralized route-management solution for autonomous vehicles in urban areas. Electronics
**2019**, 8, 722. [Google Scholar] [CrossRef][Green Version] - Babisch, W. Stress hormones in the research on cardiovascular effects of noise. Noise Health
**2003**, 5, 1. [Google Scholar] [PubMed] - World Health Organization. Environmental Noise Guidelines for the European Region; World Health Organization Regional Office for Europe UN City: Copenhagen Ø, Denmark, 2018. [Google Scholar]
- Banerjee, D. Research on road traffic noise and human health in India: Review of literature from 1991 to current. Noise Health
**2012**, 14, 113. [Google Scholar] - Silva, L.T.; Oliveira, I.S.; Silva, J.F. The impact of urban noise on primary schools. Perceptive evaluation and objective assessment. Appl. Acoust.
**2016**, 106, 2–9. [Google Scholar] [CrossRef] - Win, K.N.; Balalla, N.B.P.; Lwin, M.Z.; Lai, A. Noise-Induced Hearing Loss in the Police Force. Saf. Health Work
**2015**, 6, 134–138. [Google Scholar] [CrossRef] [PubMed][Green Version] - Rice, C. CEC joint research on annoyance due to impulse noise: Laboratory studies. In Proceedings of the Fourth International Congress on Noise as A Public Health Problem, Turin, Italy, 21–25 June 1983; Volume 2, pp. 1073–1084. [Google Scholar]
- Frei, P.; Mohler, E.; Röösli, M. Effect of nocturnal road traffic noise exposure and annoyance on objective and subjective sleep quality. Int. J. Hyg. Environ. Health
**2014**, 217, 188–195. [Google Scholar] [CrossRef] - Singh, D.; Nigam, S.P.; Agrawal, V.P.; Kumar, M. Modelling and Analysis of Urban Traffic Noise System Using Algebraic Graph Theoretic Approach. Acoust. Aust.
**2016**, 44, 249–261. [Google Scholar] [CrossRef] - Steele, C. A critical review of some traffic noise prediction models. Appl. Acoust.
**2001**, 62, 271–287. [Google Scholar] [CrossRef] - Garg, N.; Maji, S. A critical review of principal traffic noise models: Strategies and implications. Environ. Impact Assess. Rev.
**2014**, 46, 68–81. [Google Scholar] [CrossRef] - Quartieri, J.; Mastorakis, N.E.; Iannone, G.; Guarnaccia, C.; D’Ambrosio, S.; Troisi, A.; Lenza, T.L. A Review of Traffic Noise Predictive Models. In Recent Advances in Applied and Theoretical Mechanics, Proccedings of the Conference Applied and Theoretical Mechanics, Puerto de la Cruz, Tenerife, Spain, 14–16 December 2009; WSEAS Press: Athens, Greece, 2009; pp. 72–80. [Google Scholar]
- Guarnaccia, C. Advanced Tools for Traffic Noise Modelling and Prediction. WSEAS Trans. Syst.
**2013**, 12, 121–130. [Google Scholar] - Kumar, K.; Parida, M.; Katiyar, V.K. Road traffic noise prediction with neural networks–A review. Int. J. Optim. Control
**2012**, 2, 29–37. [Google Scholar] [CrossRef][Green Version] - Quinlan, J.R. Induction of decision trees. Mach. Learn.
**1986**, 1, 81–106. [Google Scholar] [CrossRef][Green Version] - Eliseeva, E.; Hubbard, A.E.; Tager, I.B. An application of machine learning methods to the derivation of exposure-response curves for respiratory outcomes. In U.C. Berkeley Division of Biostatistics Working Paper Series; University of California: Berkeley, CA, USA, 2013; p. 309. Available online: http://biostats.bepress.com/ucbbiostat/paper309 (accessed on 1 May 2013).
- Iannone, G.; Guarnaccia, C.; Quartieri, J. Noise Fundamental Diagram deduced by traffic dynamics. In Recent Researches in Geography, Geology, Energy, Environment and Biomedicine, Proceedings of the 4th WSEAS International Conference on EMESEG’11, Corfù Island, Greece, 14 July 2011; WSEAS Press: Athens, Greece, 2011; pp. 501–507. [Google Scholar]
- Quartieri, J.; Mastorakis, N.E.; Iannone, G.; Guarnaccia, C. Cellular automata application to traffic noise control. In Proceedings of the 12th WSEAS International Conference on Automatic Control, Modelling and Simulation (ACMOS’10), Sicily, Italy, 29–31 May 2010; pp. 299–304. [Google Scholar]
- Graziuso, G.; Mancini, S.; Guarnaccia, C. Comparison of single vehicle noise emission models in simulations and in a real case study by means of quantitative indicators. Int. J. Mech.
**2020**, 14, 198–207. [Google Scholar] [CrossRef] - Cammarata, G.; Cavalieri, S.; Fichera, A. A neural network architecture for noise prediction. Neural Netw.
**1995**, 8, 963–973. [Google Scholar] [CrossRef] - Kumar, P.; Nigam, S.P.; Kumar, N. Vehicular traffic noise modeling using artificial neural network approach. Transp. Res. Part C Emerg. Technol.
**2014**, 40, 111–122. [Google Scholar] [CrossRef] - Givargis, S.; Karimi, H. A basic neural traffic noise prediction model for Tehran’s roads. J. Environ. Manag.
**2010**, 91, 2529–2534. [Google Scholar] [CrossRef] - Gündoğdu, Ö.; Gökdağ, M.; Yüksel, F. A traffic noise prediction method based on vehicle composition using genetic algorithms. Appl. Acoust.
**2005**, 66, 799–809. [Google Scholar] [CrossRef] - Rahmani, S.; Mousavi, S.M.; Kamali, M.J. Modeling of road-traffic noise with the use of genetic algorithm. Appl. Soft Comput.
**2011**, 11, 1008–1013. [Google Scholar] [CrossRef] - Singh, D.; Upadhyay, R.; Pannu, H.S.; Leray, D. Development of an adaptive neuro fuzzy inference system based vehicular traffic noise prediction model. J. Ambient Intell. Hum. Comput.
**2021**, 12, 2685–2701. [Google Scholar] [CrossRef] - Guarnaccia, C.; Quartieri, J.; Tepedino, C. A hybrid predictive model for acoustic noise in urban areas based on time series analysis and artificial neural network. In AIP Conference Proceedings; AIP Publishing LLC: Melville, NY, USA, 2017; Volume 1836, p. 020069. [Google Scholar]
- Thakre, C.; Laxmi, V.; Vijay, R.; Killedar, D.J.; Kumar, R. Traffic noise prediction model of an Indian road: An increased scenario of vehicles and honking. Environ. Sci. Pollut. Res. Int.
**2020**, 27, 38311–38320. [Google Scholar] [CrossRef] [PubMed] - Kalaiselvi, R.; Ramachandraiah, A. Honking noise corrections for traffic noise prediction models in heterogeneous traffic conditions like India. Appl. Acoust.
**2016**, 111, 25–38. [Google Scholar] [CrossRef] - Sharma, A.; Bodhe, G.L.; Schimak, G. Development of a traffic noise prediction model for an urban environment. Noise Health
**2014**, 16, 63–67. [Google Scholar] [CrossRef] - IEC 61672-1: 2002. Electroacoustics—Sound level meters—Part 1: Specifications; International Electrotechnical Commission: Geneva, Switzerland, 2002. [Google Scholar]
- ISO 362-1: 2015. Measurement of Noise Emitted by Accelerating Road Vehicles—Engineering Method—Part 1: M and N Categories; International Organization for Standardization: Geneva, Switzerland, 2015. [Google Scholar]
- R Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2014; Available online: http://www.R-project.org/ (accessed on 3 May 2021).
- Breiman, L. Random forests. Mach. Learn.
**2001**, 45, 5–32. [Google Scholar] [CrossRef][Green Version] - Williams, G. Rattle: A Data Mining GUI for R. R J.
**2009**, 1, 45–55. [Google Scholar] [CrossRef][Green Version] - Breiman, L. Bagging Predictors. Mach. Learn.
**1996**, 24, 123–140. [Google Scholar] [CrossRef][Green Version] - Nelder, J.A.; Wedderburn, R.W.M. Generalized linear models. J. R. Stat. Soc. A
**1972**, 135, 370–384. [Google Scholar] [CrossRef] - Riedmiller, M.; Braun, H. A direct adaptive method for faster backpropagation learning: The RPROP algorithm. In Proceedings of the IEEE International Conference on Neural Networks, New Jersey, NJ, USA, 28 March–1 April 1993; pp. 586–591. [Google Scholar]
- Singh, D.; Nigam, S.P.; Agrawal, V.P.; Kumar, M. Vehicular traffic noise prediction using soft computing approach. J. Environ. Manag.
**2016**, 183, 59–66. [Google Scholar] [CrossRef] - Spiegel, M.R.; Stephens, L.J. Theory and Problems of Statistics, 3rd ed.; Schaum’s Outline Series; McGraw-Hill: New York, NY, USA, 1999. [Google Scholar]

**Figure 2.**Localization of the identified sites, respectively site 1 (

**a**), site 2 (

**b**), site 3 and 4 (

**c**), site 5 (

**d**).

**Figure 3.**Scatter plot of traffic volume (Q), heavy vehicles percentage (P), honking occurrences (H) and measured continuous equivalent levels (L

_{eq}), with bivariate scatter plots below the diagonal, histograms on the diagonal and the Pearson correlation above the diagonal.

**Figure 4.**Scatter plot of the models’ predictions versus measured equivalent levels in the training dataset. The black solid line is the bisector and is used as a guide for the eye, to easily identify overestimation and underestimation of the models.

**Figure 6.**A ten-fold cross validation for r, R

^{2}, MSE and accuracy for the generalized linear model (GLM).

**Table 1.**Resume of the main statistics of data and parameters used in the calibration and testing of the models. All the features have been measured with 15 min time span.

Mean | Standard Deviation | Median | Min | Max | |
---|---|---|---|---|---|

Traffic volume in 15 min (Q) [veh] | 633.8 | 341.2 | 583 | 216 | 1377 |

Percentage of heavy vehicles (P) [%] | 1.36 | 1.42 | 0.5 | 0 | 3.8 |

Total honking occurrences in 15 min (H) [counts] | 218.0 | 88.9 | 227 | 90 | 406 |

Equivalent continuous sound level (L_{eq}) [dBA] | 74.7 | 3.4 | 75 | 68 | 79.7 |

**Table 2.**Machine Learning Methods Used [41].

Model | Method | Name of Module | Input Parameters and Values |
---|---|---|---|

Decision Trees | rpart | Rpart | MinSplit = 20, MaxDepth = 30, MinBucket = 7 |

Random Forests | rf | randomForest | ntree = 500, sampling = bagging |

Linear Models | lm | glm | None |

Neural Networks | neuralnet | Neuralnet | hlayers = 10, MaxNWts = 10,000, maxit = 100 |

Machine Learning Method | r | R^{2} | MSE [dBA] | Accuracy (±1 dBA) [%] |
---|---|---|---|---|

Decision Trees (DT) | 0.935 | 0.876 | 0.884 | 68 |

Random Forests (RF) | 0.987 | 0.975 | 0.413 | 94 |

Generalized Linear Model (GLM) | 0.974 | 0.949 | 0.616 | 82 |

Neural Networks (ANN) | 0.974 | 0.949 | 0.616 | 82 |

Machine Learning Method | r | R^{2} | MSE [dBA] | Accuracy (±1.0 dBA) [%] |
---|---|---|---|---|

Decision Trees (DT) | 0.797 | 0.636 | 1.780 | 40.0 |

Random Forests (RF) | 0.955 | 0.913 | 0.797 | 60.0 |

Generalized Linear Model (GLM) | 0.965 | 0.933 | 0.666 | 80.0 |

Neural Networks (ANN) | 0.959 | 0.921 | 0.717 | 73.3 |

**Table 5.**Comparison of results for the prediction models in the training phase (without considering honking).

Machine Learning Method | r | R^{2} | MSE [dBA] | Accuracy (±1 dBA) [%] |
---|---|---|---|---|

Decision Trees (DT) | 0.933 | 0.871 | 0.902 | 64 |

Random Forests (RF) | 0.978 | 0.957 | 0.526 | 82 |

Generalized Linear Model (GLM) | 0.904 | 0.818 | 1.182 | 40 |

Neural Networks (ANN) | 0.904 | 0.818 | 1.182 | 40 |

**Table 6.**Comparison of results for the prediction models (testing dataset) without considering honking.

Machine Learning method | r | R^{2} | MSE [dBA] | Accuracy (±1.0 dBA) [%] |
---|---|---|---|---|

Decision Trees (DT) | 0.798 | 0.637 | 1.781 | 40.0 |

Random Forests (RF) | 0.955 | 0.913 | 0.809 | 66.7 |

Generalized Linear Model (GLM) | 0.880 | 0.774 | 1.291 | 46.7 |

Neural Networks (ANN) | 0.880 | 0.774 | 1.291 | 46.7 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Singh, D.; Francavilla, A.B.; Mancini, S.; Guarnaccia, C. Application of Machine Learning to Include Honking Effect in Vehicular Traffic Noise Prediction. *Appl. Sci.* **2021**, *11*, 6030.
https://doi.org/10.3390/app11136030

**AMA Style**

Singh D, Francavilla AB, Mancini S, Guarnaccia C. Application of Machine Learning to Include Honking Effect in Vehicular Traffic Noise Prediction. *Applied Sciences*. 2021; 11(13):6030.
https://doi.org/10.3390/app11136030

**Chicago/Turabian Style**

Singh, Daljeet, Antonella B. Francavilla, Simona Mancini, and Claudio Guarnaccia. 2021. "Application of Machine Learning to Include Honking Effect in Vehicular Traffic Noise Prediction" *Applied Sciences* 11, no. 13: 6030.
https://doi.org/10.3390/app11136030