Study On Urban Air Quality Assessment Based On Random Forest

Air pollution is getting more and more serious as the economy developing rapidly,the urban population growing fast and the degree of industrialization reaching higher level.How to strengthen the prevention of air pollution and pollution incidents timely and effectively has been an important issue that we are increasingly concerned about.In order to prevent the occurrence of urban air pollution incidents and ensure the quality of urban air quality,we must make an accurate and reasonable air quality evaluation and raise some efficiently prevention and control measures for emergency situations so that to offer residents a healthy living environment as far as possible.Therefore,scientific and effective air quality assessment methods play a very important role in ensuring urban air quality.However,we still widely apply traditional air quality assessment methods in China.Traditional air quality assessment methods generally consider single pollution factor and get the results from a fixed formula,furthermore,it can always blend subjective factors.With the advancement of big data and artificial intelligence,traditional methods have emerged its shortcomings of low efficiency when dealing with huge amount of data.Currently,how to make better use of big data and artificial intelligence for air quality evaluation has become a research hotspot of scholars and experts.Machine learning is one way to realize artificial intelligence.The random forest algorithm is one of machine learning algorithms that its advantages of high prediction accuracy,fast processing efficiency,strong generalization ability,and hard to over-fitting has been used by many scholars in many fields,including image classification,fault diagnosis,traffic flow prediction and other fields.Therefore,this paper uses random forest to evaluate the air quality of the city and establishes an air quality assessment model based on random forest by improving and training the model which improves the scientificity of the evaluation method.This paper establishes an air quality assessment model based on urban air quality.First,we do a lot of research on the selection of evaluation factors,evaluation criteria and existing air quality assessment methods.And then starting from the theoretical basis of random forest algorithm,we introduce and study the algorithm construction processes and its optimization method which paves the way for the establishment of the air quality evaluation model.Next,we select air quality data of 113 important environmental protection cities in China from 2014 to 2016.After the data preprocessing and data set are divided,we establish the air quality assessment model based on random forest.In the process of unbalanced data processing,node splitting and parameter adjustment,the model is further optimized and be evaluated by the accuracy and AUC value.Finally,comparing the accuracy and AUC value of the random forest algorithm,artificial neural network and support vector machine,the experimental results show that the random forest algorithm has the best evaluation effect which implies that it can accurately and effectively evaluate the air quality of urban environment.At the same time,we sort the importance of the air quality indicators by the index of out to bag and propose suggestions for future air pollution prevention and control.