# Complex Real-Time Monitoring and Decision-Making Assistance System Based on Hybrid Forecasting Module and Social Network Analysis

^{*}

## Abstract

**:**

## 1. Introduction

_{2.5}is on the rise globally and poses a serious threat to people’s health, including respiratory infections, pneumonia and lung cancer [4,5,6]. In particular, air pollution caused by crop residue burning in Northeast China is more intense than that of other regions in recent years [7]. The Air Quality Index (AQI) is used by many developed and developing countries around the world to assess air quality, considering the composition of particulate matter, gaseous pollutants and other factors. A high AQI indicates that people’s health is at risk and that government policies are needed to improve air quality [8]. Therefore, the real-time monitoring and prediction of air quality is an important basis for promoting the sustainable development of a country.

_{2.5}), particulate matter 10 (PM

_{10}), carbon dioxide (CO

_{2}), sulfur oxides (SO

_{X}), nitrogen oxides (NO

_{X}), ozone (O3), and ammonia (NH

_{3}) are exposed to the atmosphere from these activities, thus contributing to the creation of climate extremes. Such negative influences include global warming, acid rain, smog, and aerosols. As a result, air quality research has moved away from single-variable predictions to a combination of factors that increase the interpretability of the predictions [12,13,14,15].

**The main contributions of this paper could be summarized as follows:**

- (1)
- Text sentiment analysis is performed to explore public emotions related to air quality, which is then introduced to the construct of explanatory variables. It is verified that adding public emotions improves the performance of the forecasting model.
- (2)
- A feature processing strategy based on multiple feature selection methods and result fusion is innovatively proposed to solve the problem of difficulty in extracting features from air pollution data.
- (3)
- A CNN-D-LSTM is constructed by adding a DenseNet, which greatly reduces the probability of parameter explosion and improves the ability to extract useful information automatically, thus contributing to the superiority of forecasting performance.
- (4)
- Social network analysis is introduced to improve the interpretability of air pollution correlations in urban agglomerations. Moreover, the additional social analysis is conducive to dynamic monitoring and timely policy-making.
- (5)
- The combination of forecasting and social analysis could be expanded to many other fields for helping the exploring of cluster change and other applications, which is also an advancement of spatial correlation analysis.

## 2. Methodology

#### 2.1. Problems and Motivations

#### 2.2. Text Sentiment Analysis

**Step 1:**First, identify the mainstream platforms or forums that are geared towards this based on the volume of users, and then utilize crawling techniques to obtain comments on air quality from these platforms.

**Step 2:**Jieba’s word separation algorithm was utilized in text information preprocessing, including deactivation and text vectorization. In this case, the implementation of the word separation algorithm is performed as follows:

**Step 3:**The word vectors were then subjected to feature extraction and sentiment classification to identify keywords that reflect public emotions.

**Step 4:**Based on the above keywords, a Baidu search index corresponding to the date that the air pollution data were obtained and used as a reflection of public sentiment. Respectively, the Baidu index includes both computer and mobile.

#### 2.3. Feature Processing

**(1) Filter algorithm:**To generate effective influencing factor subsets for air quality forecasting, it is important to filter less crucial features. On one hand, appropriate feature filtering can effectively avoid the dimension explosion problem in the subsequent substitution of machine learning models, which is conducive to improving model adaptability. On the other hand, in the case of different urban agglomerations, there may be differences in the factors influencing air quality, and adaptive filtering can help to find the key influencing factors. In this study, the grey correlation analysis served as a filter algorithm to eliminate variables of lower importance, while significant features were selected. The corresponding formula is as follows:

**(2) Embedded algorithm:**After the feature subsets are acquired, it is essential to evaluate these features from another perspective. In this paper, LASSO was utilized as an embedded algorithm. LASSO obtains a more refined model by constructing a penalty function such that it compresses some of the regression coefficients. Moreover, it forces an absolute sum of the coefficients to be less than some fixed value, while it sets some of the regression coefficients to zero [51].

**(3) Result fusion:**Principal component analysis (PCA) is known as a classic method for high-dimensional data preparation, especially in the field of explanatory data analysis and forecasting model conducting [53]. It specializes in data degradation, which not only preserves key information but also removes unanticipated noise [54]. The PCA algorithm is executed as shown in Table 1.

#### 2.4. Forecasting Module

#### 2.4.1. CNN Layer

#### 2.4.2. LSTM and Output Layer

#### 2.5. Social Network Analysis

#### 2.6. Evaluation Matrix

## 3. Case Study

#### 3.1. Study Area and Data Description

_{2.5}, PM

_{10}, SO

_{2}, NO

_{2}, CO and O

_{3}; (2) Meteorological data: Cumulative daily precipitation, cumulative daily light, average air temperature, average air pressure, average wind speed, and average humidity; (3) Public emotions: Haze Index and Environmental Pollution Index, including mobile, computer and total indices. The length of these variables is the same as that of AQI. Air pollutant concentrations are collected from the website http://data.cma.cn (accessed on16 May 2023) while the data representing public emotions are from the website https://index.baidu.com (accessed on16 May 2023).

#### 3.2. Data Preprocessing

#### 3.3. The Simulation Results of Forecasting Module

#### 3.3.1. Compare of Single Model in Urban Agglomeration Forecasting

**Remark**

**1:**

#### 3.3.2. Compare the Performance on Different Clusters Divided

**Remark**

**2:**

#### 3.4. The Properties Analysis of Network

**Remark**

**3:**

## 4. Discussion

#### 4.1. The Dynamic Analysis of Social Network

#### 4.2. The Stability of the Proposed System

#### 4.3. Multistep Forecasting of Proposed System

## 5. Conclusions

#### 5.1. Main Conclusions

#### 5.2. Academic Significance

#### 5.3. Practical Application

#### 5.4. Future Research Directions

## Supplementary Materials

## Author Contributions

## Funding

## Data Availability Statement

## Conflicts of Interest

## References

- Miao, S.; Gangolells, M.; Tejedor, B. Data-Driven Model for Predicting Indoor Air Quality and Thermal Comfort Levels in Naturally Ventilated Educational Buildings Using Easily Accessible Data for Schools. J. Build. Eng.
**2023**, 80, 108001. [Google Scholar] [CrossRef] - Manisalidis, I.; Stavropoulou, E.; Stavropoulos, A.; Bezirtzoglou, E. Environmental and Health Impacts of Air Pollution: A Review. Front. Public Health
**2020**, 8, 505570. [Google Scholar] [CrossRef] [PubMed] - Lee, H.; Hwang-Bo, H.; Ji, S.Y.; Kim, M.Y.; Kim, S.Y.; Park, C.; Hong, S.H.; Kim, G.-Y.; Song, K.S.; Hyun, J.W.; et al. Diesel Particulate Matter2.5 Promotes Epithelial-Mesenchymal Transition of Human Retinal Pigment Epithelial Cells via Generation of Reactive Oxygen Species. Environ. Pollut.
**2020**, 262, 114301. [Google Scholar] [CrossRef] [PubMed] - Phruksahiran, N. Improvement of Air Quality Index Prediction Using Geographically Weighted Predictor Methodology. Urban Clim.
**2021**, 38, 100890. [Google Scholar] [CrossRef] - Kypreos, S.; Glynn, J.; Panos, E.; Giannakidis, G.; Ó Gallachóir, B. Efficient and Equitable Climate Change Policies. Systems
**2018**, 6, 10. [Google Scholar] [CrossRef] - Shah, S.A.R.; Zhang, Q.; Abbas, J.; Balsalobre-Lorente, D.; Pilař, L. Technology, Urbanization and Natural Gas Supply Matter for Carbon Neutrality: A New Evidence of Environmental Sustainability under the Prism of COP26. Resour. Policy
**2023**, 82, 103465. [Google Scholar] [CrossRef] - Yin, S.; Wang, X.; Zhang, X.; Zhang, Z.; Xiao, Y.; Tani, H.; Sun, Z. Exploring the effects of crop residue burning on local haze pollution in Northeast China using ground and satellite data. Atmos. Environ.
**2019**, 199, 189–201. [Google Scholar] [CrossRef] - Li, H.; Wang, J.; Li, R.; Lu, H. Novel Analysis-Forecast System Based on Multi-Objective Optimization for Air Quality Index. J. Clean. Prod.
**2019**, 208, 1365–1383. [Google Scholar] [CrossRef] - Ravindra, K.; Singh, T.; Pandey, V.; Mor, S. Air Pollution Trend in Chandigarh City Situated in Indo-Gangetic Plains: Understanding Seasonality and Impact of Mitigation Strategies. Sci. Total Environ.
**2020**, 729, 138717. [Google Scholar] [CrossRef] - Ravindra, K. Emission of Black Carbon from Rural Households Kitchens and Assessment of Lifetime Excess Cancer Risk in Villages of North India. Environ. Int.
**2019**, 122, 201–212. [Google Scholar] [CrossRef] - Rupakheti, D.; Kim Oanh, N.T.; Rupakheti, M.; Sharma, R.K.; Panday, A.K.; Puppala, S.P.; Lawrence, M.G. Indoor Levels of Black Carbon and Particulate Matters in Relation to Cooking Activities Using Different Cook Stove-Fuels in Rural Nepal. Energy Sustain. Dev.
**2019**, 48, 25–33. [Google Scholar] [CrossRef] - Zhang, C.; Ran, L.; Song, L. Fast Alignment of SINS for Marching Vehicles Based on Multi-Vectors of Velocity Aided by GPS and Odometer. Sensors
**2018**, 18, 137. [Google Scholar] [CrossRef] [PubMed] - Chen, L.; Ding, Y.; Lyu, D.; Liu, X.; Long, H. Deep Multi-Task Learning Based Urban Air Quality Index Modelling. Proc. ACM Interact. Mob. Wearable Ubiquitous Technol.
**2019**, 3, 2:1–2:17. [Google Scholar] [CrossRef] - Rowley, A.; Karakuş, O. Predicting Air Quality via Multimodal AI and Satellite Imagery. Remote Sens. Environ.
**2023**, 293, 113609. [Google Scholar] [CrossRef] - Liu, J.; Li, D.; Shan, W.; Liu, S. A Feature Selection Method Based on Multiple Feature Subsets Extraction and Result Fusion for Improving Classification Performance. Appl. Soft Comput.
**2023**, 150, 111018. [Google Scholar] [CrossRef] - Xu, H.; Li, X.; Yin, J.; Zhou, L.; Song, Y.; Hu, T. Microphysics Affect the Sensitivities of Rainfall to Different Horizontal-Resolution Simulations: Evidence from a Case Study of the Weather Research and Forecasting Model Runs. Atmos. Res.
**2023**, 296, 107022. [Google Scholar] [CrossRef] - Zhang, C.; Jing, D.; Wu, C.; Li, S.; Cheng, N.; Li, W.; Wang, G.; Chen, B.; Wang, Q.; Hu, J. Integrating Chemical Mass Balance and the Community Multiscale Air Quality Models for Source Identification and Apportionment of PM2.5. Process Saf. Environ. Prot.
**2021**, 149, 665–675. [Google Scholar] [CrossRef] - Jurado, X.; Reiminger, N.; Vazquez, J.; Wemmert, C.; Dufresne, M.; Blond, N.; Wertel, J. Assessment of Mean Annual NO2 Concentration Based on a Partial Dataset. Atmos. Environ.
**2020**, 221, 117087. [Google Scholar] [CrossRef] - Yoo, E.-H.; Zammit-Mangion, A.; Chipeta, M.G. Adaptive Spatial Sampling Design for Environmental Field Prediction Using Low-Cost Sensing Technologies. Atmos. Environ.
**2020**, 221, 117091. [Google Scholar] [CrossRef] - Ravindiran, G.; Hayder, G.; Kanagarathinam, K.; Alagumalai, A.; Sonne, C. Air Quality Prediction by Machine Learning Models: A Predictive Study on the Indian Coastal City of Visakhapatnam. Chemosphere
**2023**, 338, 139518. [Google Scholar] [CrossRef] - Sharma, E.; Deo, R.C.; Prasad, R.; Parisi, A.V. A Hybrid Air Quality Early-Warning Framework: An Hourly Forecasting Model with Online Sequential Extreme Learning Machines and Empirical Mode Decomposition Algorithms. Sci. Total Environ.
**2020**, 709, 135934. [Google Scholar] [CrossRef] [PubMed] - Ahmed, A.A.M.; Jui, S.J.J.; Sharma, E.; Ahmed, M.H.; Raj, N.; Bose, A. An Advanced Deep Learning Predictive Model for Air Quality Index Forecasting with Remote Satellite-Derived Hydro-Climatological Variables. Sci. Total Environ.
**2024**, 906, 167234. [Google Scholar] [CrossRef] - Li, B.; Wu, B.; Peng, Y.; Cai, W. Tube-Based Robust Model Predictive Control of Multi-Zone Demand-Controlled Ventilation Systems for Energy Saving and Indoor Air Quality. Appl. Energy
**2022**, 307, 118297. [Google Scholar] [CrossRef] - Chen, S.; Zheng, L. Complementary Ensemble Empirical Mode Decomposition and Independent Recurrent Neural Network Model for Predicting Air Quality Index. Appl. Soft Comput.
**2022**, 131, 109757. [Google Scholar] [CrossRef] - Prieler, R.; Mayrhofer, M.; Gaber, C.; Gerhardter, H.; Schluckner, C.; Landfahrer, M.; Eichhorn-Gruber, M.; Schwabegger, G.; Hochenauer, C. CFD-Based Optimization of a Transient Heating Process in a Natural Gas Fired Furnace Using Neural Networks and Genetic Algorithms. Appl. Therm. Eng.
**2018**, 138, 217–234. [Google Scholar] [CrossRef] - Arsov, M.; Zdravevski, E.; Lameski, P.; Corizzo, R.; Koteli, N.; Gramatikov, S.; Mitreski, K.; Trajkovik, V. Multi-Horizon Air Pollution Forecasting with Deep Neural Networks. Sensors
**2021**, 21, 1235. [Google Scholar] [CrossRef] [PubMed] - Yan, R.; Liao, J.; Yang, J.; Sun, W.; Nong, M.; Li, F. Multi-Hour and Multi-Site Air Quality Index Forecasting in Beijing Using CNN, LSTM, CNN-LSTM, and Spatiotemporal Clustering. Expert Syst. Appl.
**2021**, 169, 114513. [Google Scholar] [CrossRef] - Qi, Z.; Wang, T.; Song, G.; Hu, W.; Li, X.; Zhongfei, Z. Deep Air Learning: Interpolation, Prediction, and Feature Analysis of Fine-Grained Air Quality. IEEE Trans. Knowl. Data Eng.
**2017**, 30, 2285–2297. [Google Scholar] [CrossRef] - Wang, B.; Kong, W.; Guan, H.; Xiong, N.N. Air Quality Forecasting Based on Gated Recurrent Long Short Term Memory Model in Internet of Things. IEEE Access
**2019**, 7, 69524–69534. [Google Scholar] [CrossRef] - Wu, Z.; Luo, G.; Yang, Z.; Guo, Y.; Li, K.; Xue, Y. A comprehensive review on deep learning approaches in wind forecasting applications. CAAI Trans. Intell. Technol.
**2022**, 7, 129–143. [Google Scholar] [CrossRef] - Bahdanau, D.; Cho, K.; Bengio, Y. Neural Machine translation by jointly learning to align and translate. arXiv
**2015**, arXiv:1409.0473. [Google Scholar] - Yang, C.; Jiang, M.; Jiang, B.; Zhou, W.; Li, K. Co-Attention Network with Question Type for Visual Question Answering. IEEE Access
**2019**, 7, 40771–40781. [Google Scholar] [CrossRef] - Xia, Q.; Yu, C.; Hou, Y.; Peng, P.; Zheng, Z.; Chen, W. Multi-Modal Alignment of Visual Question Answering Based on Multi-Hop Attention Mechanism. Electronics
**2022**, 11, 1778. [Google Scholar] [CrossRef] - Osman, A.; Samek, W. DRAU: Dual Recurrent Attention Units for Visual Question Answering. Comput. Vis. Image Underst.
**2019**, 185, 24–30. [Google Scholar] [CrossRef] - Huang, Y.; Liu, S.; Yang, L. Wind Speed Forecasting Method Using EEMD and the Combination Forecasting Method Based on GPR and LSTM. Sustainability
**2018**, 10, 3693. [Google Scholar] [CrossRef] - Liu, H.; Mi, X.; Li, Y. Wind speed forecasting method based on deep learning strategy using empirical wavelet transform, long short term memory neural network and Elman neural network. Energy Convers. Manag.
**2018**, 156, 498–514. [Google Scholar] [CrossRef] - Wu, Y.-X.; Wu, Q.-B.; Zhu, J.-Q. Data-driven wind speed forecasting using deep feature extraction and LSTM. IET Renew. Power Gener.
**2019**, 13, 2062–2069. [Google Scholar] [CrossRef] - Ma, J.; Ding, Y.; Gan, V.J.L.; Lin, C.; Wan, Z. Spatiotemporal Prediction of PM2.5 Concentrations at Different Time Granularities Using IDW-BLSTM. IEEE Access
**2019**, 7, 107897–107907. [Google Scholar] [CrossRef] - Heydari, S.S.; Mountrakis, G. Meta-Analysis of Deep Neural Networks in Remote Sensing: A Comparative Study of Mono-Temporal Classification to Support Vector Machines. ISPRS J. Photogramm. Remote Sens.
**2019**, 152, 192–210. [Google Scholar] [CrossRef] - Sharma, A.; Liu, X.; Yang, X. Land Cover Classification from Multi-Temporal, Multi-Spectral Remotely Sensed Imagery Using Patch-Based Recurrent Neural Networks. Neural Netw.
**2018**, 105, 346–355. [Google Scholar] [CrossRef] - Gu, X.; Li, H.; Fan, H. Spatiotemporal Hybrid Air Pollution Early Warning System of Urban Agglomeration Based on Adaptive Feature Extraction and Hesitant Fuzzy Cognitive Maps. Systems
**2023**, 11, 286. [Google Scholar] [CrossRef] - Peralta, G.; Zareei, A. A Network Approach to Portfolio Selection. J. Empir. Financ.
**2016**, 38, 157–180. [Google Scholar] [CrossRef] - Výrost, T.; Lyócsa, Š.; Baumöhl, E. Network-Based Asset Allocation Strategies. N. Am. J. Econ. Financ.
**2019**, 47, 516–536. [Google Scholar] [CrossRef] - Wen, D.; Ma, C.; Wang, G.-J.; Wang, S. Investigating the Features of Pairs Trading Strategy: A Network Perspective on the Chinese Stock Market. Phys. A Stat. Mech. Its Appl.
**2018**, 505, 903–918. [Google Scholar] [CrossRef] - Sillesen, M.; Bambakidis, T.; Dekker, S.E.; Li, Y.; Alam, H.B. Fresh Frozen Plasma Modulates Brain Gene Expression in a Swine Model of Traumatic Brain Injury and Shock: A Network Analysis. J. Am. Coll. Surg.
**2017**, 224, 49–58. [Google Scholar] [CrossRef] [PubMed] - Dong, G.; Fan, J.; Shekhtman, L.M.; Shai, S.; Du, R.; Tian, L.; Chen, X.; Stanley, H.E.; Havlin, S. Resilience of Networks with Community Structure Behaves as If under an External Field. Proc. Natl. Acad. Sci. USA
**2018**, 115, 6911–6915. [Google Scholar] [CrossRef] - Wang, M.; Tian, L.; Du, R. Research on the Interaction Patterns among the Global Crude Oil Import Dependency Countries: A Complex Network Approach. Appl. Energy
**2016**, 180, 779–791. [Google Scholar] [CrossRef] - Du, R.; Dong, G.; Tian, L.; Wang, Y.; Zhao, L.; Zhang, X.; Vilela, A.L.M.; Stanley, H.E. Identifying the Peak Point of Systemic Risk in International Crude Oil Importing Trade. Energy
**2019**, 176, 281–291. [Google Scholar] [CrossRef] - Lin, S.-Y.; Kung, Y.-C.; Leu, F.-Y. Predictive Intelligence in Harmful News Identification by BERT-Based Ensemble Learning Model with Text Sentiment Analysis. Inf. Process. Manag.
**2022**, 59, 102872. [Google Scholar] [CrossRef] - Liu, L.; Fu, Q.; Lu, Y.; Wang, Y.; Wu, H.; Chen, J. CorrDQN-FS: A Two-Stage Feature Selection Method for Energy Consumption Prediction via Deep Reinforcement Learning. J. Build. Eng.
**2023**, 80, 108044. [Google Scholar] [CrossRef] - Cui, L.; Bai, L.; Wang, Y.; Yu, P.S.; Hancock, E.R. Fused Lasso for Feature Selection Using Structural Information. Pattern Recognit.
**2021**, 119, 108058. [Google Scholar] [CrossRef] - Mozafari, Z.; Chamjangali, M.A.; Arashi, M. Combination of Least Absolute Shrinkage and Selection Operator with Bayesian Regularization Artificial Neural Network (LASSO-BR-ANN) for QSAR Studies Using Functional Group and Molecular Docking Mixed Descriptors-ScienceDirect. Chemom. Intell. Lab. Syst.
**2020**, 200, 103998. [Google Scholar] [CrossRef] - Ouassila, B.; Zohra, T.F.; Laid, L.; Hizia, B. Neural Networks Based Linear (PCA) and Nonlinear (ISOMAP) Feature Extraction for Soil Swelling Pressure Prediction (North East Algeria). Heliyon
**2023**, 9, e18673. [Google Scholar] [CrossRef] [PubMed] - Zhao, K.; Xiao, J.; Li, C.; Xu, Z.; Yue, M. Fault Diagnosis of Rolling Bearing Using CNN and PCA Fractal Based Feature Extraction. Measurement
**2023**, 223, 113754. [Google Scholar] [CrossRef] - Yu, Y.; Si, X.; Hu, C.; Zhang, J. A Review of Recurrent Neural Networks: LSTM Cells and Network Architectures. Neural Comput.
**2019**, 31, 1235–1270. [Google Scholar] [CrossRef] [PubMed] - Skrobek, D.; Krzywanski, J.; Sosnowski, M.; Kulakowska, A.; Zylka, A.; Grabowska, K.; Ciesielska, K.; Nowak, W. Implementation of deep learning methods in prediction of adsorption processes. Adv. Eng. Softw.
**2022**, 173, 103190. [Google Scholar] [CrossRef] - Ahmed, I.; Ahmad, M.; Chehri, A.; Jeon, G. A heterogeneous network embedded medicine recommendation system based on LSTM. Future Gener. Comput. Syst.
**2023**, 149, 1–11. [Google Scholar] [CrossRef] - Latifi, M.; Kerachian, R.; Beig Zali, R. Evaluating Energy Harvesting from Water Distribution Networks Using Combined Stakeholder and Social Network Analysis. Energy Strategy Rev.
**2023**, 49, 101158. [Google Scholar] [CrossRef] - He, X.; Zhao, K.; Chu, X. AutoML: A survey of the state-of-the-art. Knowl.-Based Syst.
**2021**, 212, 106622. [Google Scholar] [CrossRef] - Krzywanski, J.; Skrobek, D.; Zylka, A.; Grabowska, K.; Kulakowska, A.; Sosnowski, M.; Nowak, W.; Blanco-Marigorta, A. Heat and mass transfer prediction in fluidized beds of cooling and desalination systems by AI approach. Appl. Therm. Eng.
**2023**, 225, 120200. [Google Scholar] [CrossRef]

Algorithm Ⅰ: data preprocessing |

1 /* Detect the abnormal value */ |

2 /* Calculate the local median and standard deviation $\sigma $ of time series */ |

3 /* Set the initial value of window length k and threshold $\kappa $*/ |

4 If $\left|{x}_{i-k}-\left|{\overrightarrow{E}}_{m}^{i}\right|\right|>\kappa \sigma (i,k)$; |

5 where $\left|{\overrightarrow{E}}_{m}^{i}\right|$ represents the value calculated by Hampel Filter |

6 /* remove the raw value with $\left|{\overrightarrow{E}}_{m}^{i}\right|$ */ |

7 End |

Algorithm Ⅱ: feature preprocessing |

8 /* Input the explanatory variables */ |

9 /* Calculate the grey correlations between each one with respond variable */ |

10 /* Sort the explanatory variables based on the absolute value of grey correlations */ |

11 /* Initialize the number of chosen variables n */ |

12 If the rank of variable is lower than n; |

13 This variable would be removed |

14 Else if the rank is higher than n; |

15 This variable would be selected and put into $\Im $ |

16 End |

17 /* Generate the multiple regression corresponding to each variable in $\Im $*/ |

18 /* Add penalty function to certain variable */ |

19 /* Record each influencing factor to the respond variable */ |

20 /* Select the most important variable from $\Im $*/ |

21 /* Set the initial value of factors number after dimension */ |

22 /* Compute the normalized feature vector */ |

23
$\mathrm{\Phi}=\frac{1}{n}{\displaystyle \sum _{k=1}^{n}{x}_{k}}$ |

24 where denotes the feature vector, and n is the total number. |

25 /* Calculate the covariance matrix */ |

26 $\mathrm{\Lambda}=\frac{1}{n}{\displaystyle \sum _{k=1}^{n}({x}_{k}-\mathrm{\Phi}){({x}_{k}-\mathrm{\Phi})}^{T}}$ |

27 /* Solve the eigen value */ |

28
${\pi}_{i}={\lambda}_{i}{\upsilon}_{i}$ |

29 where ${\lambda}_{i}$ and ${\upsilon}_{i}$ represent the eigen values and vectors of covariance matrix. |

30 /* Estimate the high-valued eigen vectors */ |

31 /* Sort all eigenvalues in descending order */ |

32 /* Set the threshold value $\theta $*/ |

33 /* Select high-valued eigen ${\lambda}_{i}$ based on the following principles */ |

34
$({\displaystyle \sum _{i=1}^{s}{\lambda}_{i}}){({\displaystyle \sum _{i=1}^{s}{\lambda}_{i}})}^{-1}\ge \theta $ |

35 where s donates the number of ${\lambda}_{i}$ selected. |

36 /* Select eigen vectors corresponding to ${\lambda}_{i}$*/ |

Algorithm Ⅲ: Forecasting Module |

Input: the respond AQI series after Hampel Filter |

the explanatory time vector after feature selection and result fusion |

Output: MAE, RMSE, SMAPE, U_{1,} r |

Parameters: Number of hidden units |

Max epochs |

Initial learning rate |

Learning rate drop factor |

37 /* CNN layer capture the features along time */ |

38 /* LSTM layer remember information of the last time and forget the useless one */ |

39 /* DenseNet deal with the information coming from each direction */ |

40 /* Output layer combine the above and gain the output */ |

City | Observations | Mean | Std | Min | Max | Skewness | Kurtosis |
---|---|---|---|---|---|---|---|

Harbin | 2147 | 77.2128 | 60.1324 | 9 | 466 | 2.4236 | 10.4447 |

Daqing | 2147 | 57.0326 | 40.1298 | 11 | 478 | 3.2356 | 19.2091 |

Qiqihar | 2147 | 58.3456 | 39.5789 | 9 | 385 | 3.2705 | 18.4583 |

Suihua | 2147 | 60.0466 | 49.6613 | 8 | 491 | 3.1722 | 16.8502 |

Mudanjiang | 2147 | 60.1723 | 32.8704 | 12 | 327 | 2.2973 | 12.0391 |

Changchun | 2147 | 74.0703 | 46.8634 | 10 | 425 | 2.3526 | 10.7278 |

Jilin | 2147 | 70.8337 | 45.1052 | 11 | 401 | 2.3928 | 11.1166 |

Siping | 2147 | 73.7038 | 41.9471 | 9 | 485 | 2.4828 | 14.1539 |

Liaoyuan | 2147 | 42.4965 | 34.7778 | 3 | 372 | 2.5200 | 13.8902 |

Songyuan | 2147 | 33.5706 | 37.1550 | 3 | 537 | 4.6748 | 41.4423 |

Yanbian | 2143 | 51.7653 | 29.7142 | 12 | 292 | 2.4935 | 13.0742 |

${\mathit{\epsilon}}_{\mathit{M}\mathit{A}\mathit{E}}$ | ${\mathit{\epsilon}}_{\mathit{R}\mathit{M}\mathit{S}\mathit{E}}$ | ${\mathit{\epsilon}}_{\mathit{S}\mathit{M}\mathit{A}\mathit{P}\mathit{E}}$ | ${\mathit{\epsilon}}_{\mathit{U}1}$ | ${\mathit{\epsilon}}_{\mathit{r}}$ | |
---|---|---|---|---|---|

BP | 24.2013 | 36.4840 | 46.1941 | 0.2765 | 0.6357 |

ELMAN | 21.6052 | 30.8333 | 38.6090 | 0.2335 | 0.6430 |

GM(1,n) | 57.6189 | 73.7039 | 70.4562 | 0.3597 | 0.9475 |

LSSVM | 15.8119 | 27.6802 | 29.3036 | 0.1888 | 0.8012 |

RF | 12.3909 | 20.9718 | 20.6981 | 0.1376 | 0.8808 |

ARIMAX | 12.0074 | 17.6970 | 19.6029 | 0.1199 | 0.9696 |

LSTM | 16.1042 | 20.4481 | 27.2522 | 0.1629 | 0.8180 |

GRU | 15.6653 | 20.2217 | 26.9076 | 0.1614 | 0.8106 |

CNN-D-LSTM | 7.8692 | 9.9289 | 11.5215 | 0.0744 | 0.9816 |

${\mathit{\epsilon}}_{\mathit{M}\mathit{A}\mathit{E}}$ | ${\mathit{\epsilon}}_{\mathit{R}\mathit{M}\mathit{S}\mathit{E}}$ | ${\mathit{\epsilon}}_{\mathit{S}\mathit{M}\mathit{A}\mathit{P}\mathit{E}}$ | ${\mathit{\epsilon}}_{\mathit{U}1}$ | ${\mathit{\epsilon}}_{\mathit{r}}$ | |||
---|---|---|---|---|---|---|---|

K-mean clustering | |||||||

Contour coefficient | 0.0707 | Cluster I | 7.5654 | 10.1764 | 9.5358 | 0.0643 | 0.9844 |

CH Index | 2.3172 | Cluster II | 5.5383 | 6.8217 | 8.2349 | 0.0603 | 0.9872 |

DB Index | 1.4069 | Cluster III | 7.8209 | 9.8334 | 10.103 | 0.0700 | 0.9794 |

Hierarchical clustering | |||||||

Contour coefficient | 0.2202 | Cluster I | 7.5654 | 10.1764 | 9.5358 | 0.0643 | 0.9844 |

CH Index | 3.5763 | Cluster II | 6.9235 | 8.9021 | 9.4959 | 0.0661 | 0.9740 |

DB Index | 1.2381 | Cluster III | 6.0123 | 8.7131 | 9.9058 | 0.0633 | 0.9808 |

Gaussian hybrid clustering | |||||||

Contour coefficient | 0.2200 | Cluster I | 6.9638 | 9.0379 | 7.9881 | 0.0570 | 0.9886 |

CH Index | 3.8193 | Cluster II | 4.2317 | 6.1967 | 6.8370 | 0.0535 | 0.9824 |

DB Index | 0.7820 | Cluster III | 5.0357 | 7.2345 | 8.3843 | 0.0596 | 0.9812 |

Out-Degree | In-Degree | Degree | Betweenness | Closeness | |
---|---|---|---|---|---|

Harbin | 7 | 6 | 8 | 4.5 | 23 |

Daqing | 7 | 7 | 10 | 11.4 | 21 |

Qiqihar | 8 | 6 | 9 | 8 | 22 |

Suihua | 0 | 0 | 5 | 1.5 | 26 |

Mudanjiang | 2 | 3 | 4 | 0.4 | 27 |

Changchun | 2 | 3 | 3 | 0 | 28 |

Jilin | 0 | 0 | 2 | 0 | 29 |

Siping | 3 | 3 | 3 | 0 | 28 |

Liaoyuan | 3 | 3 | 4 | 0.4 | 27 |

Songyuan | 3 | 3 | 4 | 0.4 | 27 |

Yanbian | 2 | 3 | 4 | 0.4 | 27 |

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |

© 2024 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**

Fan, H.; Li, H.; Gu, X.; Ren, Z.
Complex Real-Time Monitoring and Decision-Making Assistance System Based on Hybrid Forecasting Module and Social Network Analysis. *Systems* **2024**, *12*, 39.
https://doi.org/10.3390/systems12020039

**AMA Style**

Fan H, Li H, Gu X, Ren Z.
Complex Real-Time Monitoring and Decision-Making Assistance System Based on Hybrid Forecasting Module and Social Network Analysis. *Systems*. 2024; 12(2):39.
https://doi.org/10.3390/systems12020039

**Chicago/Turabian Style**

Fan, Henghao, Hongmin Li, Xiaoyang Gu, and Zhongqiu Ren.
2024. "Complex Real-Time Monitoring and Decision-Making Assistance System Based on Hybrid Forecasting Module and Social Network Analysis" *Systems* 12, no. 2: 39.
https://doi.org/10.3390/systems12020039