R-LSTM Algorithm for Imputation of Air Quality Data Based on Sample Screening

In the context of increasing societal concern for the natural environment, urban air quality monitoring has become a key component of environmental management. However, due to technical failures of monitoring equipment and data transmission problems, the lack of air quality data often affects the analysis and progress of downstream tasks. To solve this problem, this paper proposes an air quality data interpolation algorithm based on the Relevance Long Short Term Memory Network (R-LSTM). The algorithm significantly improves the accuracy of missing data interpolation by exploiting the intrinsic correlation between air quality indicators and combining it with a sample screening mechanism. In the experimental part, this paper uses air quality time series data from Shenyang City, Liaoning Province for testing. Under the set condition of 15 % random missing rate, this paper compares the performance of R-LSTM with other benchmark models using the mean absolute error (MAE) and mean relative error (MRE) as evaluation indices. The experimen-tal results show that R-LSTM has better results in interpolating the missing data of air quality.