Study On Public Environmental Emotion Prediction Model

In recent years,with the continuous improvement of the public’s environmental behavior and awareness,the public’s demands and expectations for the environment are also on the rise.The change of public environmental emotions caused by environmental problems may cause social group anxiety and irritability.So,it is of great significance to analyze the influencing factors of public environmental emotion and establish a prediction model of public environmental emotion to prevent social group events caused by environmental problems in advance.Public environmental emotion belong to psychological spatial information,so how to quantify the change of public environmental emotion caused by the change of environmental quality,comprehensively analyze the atmospheric physical and chemical factors that affect the public’s environmental mood,and realize the quantification and predictability of public environmental emotion is a difficult problem of the field of public environmental research.Based on the Public Participatory Sensing,this paper applies the data onto“Public Environmental Satisfaction”collected by“Impression Ecology”APP to analyze the relationship between the atmospheric environment and public environmental emotion,and proposes a prediction model of public environmental emotion.The details of the researches are listed as follows:Firstly,analysis of influencing factors of public environmental emotion.Taking the data of Lanzhou as an example,the OLS model was established based on atmospheric environmental factors and public environmental satisfaction.On the basis of OSL backward elimination procedure,screen the PM_2.5,PM₁₀,Temperature（TMP）,Humidity（HUM）as the critical ones from atmospheric environmental factors and the Pearson correlation analysis was conducted with public environmental satisfaction.The experimental results show that there is a strong negative correlation（-0.77 and-0.61respectively）with public environmental satisfaction between PM_2.5 and PM₁₀.However,there was a weak positive correlation between TMP and public environmental satisfaction as well as HUM（0.25 and 0.15 respectively）.Therefore,reducing the concentration of PM_2.5 and PM₁₀ in Lanzhou has a positive effect on improving the public’s environmental satisfaction.Secondly,dynamic relationship analysis of public environmental sentiment and its influencing factors in time series.Construct a vector auto regressive model（VAR）of public environmental emotion and its time series influencing factors PM_2.5,PM₁₀,TMP and HUM.According to the stationarity test of the variable system,the lag order is calculated according to the characteristics of the model,and through the AR characteristic root test,the VAR（4）model of the 4-order lag phase is determined to be constructed.Granger causality test,generalized impulse response analysis and variance decomposition were used to analyze the dynamic effects of PM_2.5,PM₁₀,TMP and HUM on public environmental sentiment.The results show that when the model lag phase is 4,the impact of PM₁₀ and TMP on public environmental satisfaction is positively correlated in the early stage and negatively correlated in the later stage.The overall impact of PM_2.5 on public environmental satisfaction is negatively correlated,and the overall impact of HUM on public environmental satisfaction is positively correlated.Therefore,to improve the public environmental sentiment in Lanzhou,reducing PM_2.5 and increase HUM is the primary path.Finally,construction of the prediction model of public environmental emotion.Based on LSTM and RNN,the public environment emotion prediction model was built,and analyze the advantages of LSTM in the prediction of public environmental sentiment.The results show that compared with RNN,the public environmental emotion prediction model based on LSTM does well in the learning process.The prediction accuracy of LSTM（90.53%）is 3.58%higher than RNN（86.95%）in the prediction of the public mood in environment.The mean error of LSTM（1.0731%）is 0.1385%lower than RNN（1.2116%）in the prediction of the public mood in environment.Therefore,the public environmental emotion prediction model based on LSTM can predict the public environmental emotion prediction level through the physical spatial information with good effect.