Long-term trends of wet inorganic nitrogen deposition in Rocky Mountain National Park: Influence of missing data imputation methods and associated uncertainty

AbstractGraphical AbstractLand use regression (LUR) has been widely applied in epidemiologic research for exposure assessment. In this study, for the first time, we aimed to develop a spatiotemporal LUR model using Distributed Space Time Expectation Maximization (D-STEM). This spatiotemporal LUR model examined with daily particulate matter ≤ 2.5 μm (PM2.5) within the megacity of Tehran, capital of Iran. Moreover, D-STEM missing data imputation was compared with mean substitution in each monitoring station, as it is equivalent to ignoring of missing data, which is common in LUR studies that employ regulatory monitoring stations’ data. The amount of missing data was 28% of the total number of observations, in Tehran in 2015. The annual mean of PM2.5 concentrations was 33 μg/m3. Spatiotemporal R-squared of the D-STEM final daily LUR model was 78%, and leave-one-out cross-validation (LOOCV) R-squared was 66%. Spatial R-squared and LOOCV R-squared were 89% and 72%, respectively. Temporal R-squared and LOOCV R-squared were 99.5% and 99.3%, respectively. Mean absolute error decreased 26% in imputation of missing data by using the D-STEM final LUR model instead of mean substitution. This study reveals competence of the D-STEM software in spatiotemporal missing data imputation, estimation of temporal trend, and mapping of small scale (20 × 20 meters) within-city spatial variations, in the LUR context. The estimated PM2.5 concentrations maps could be used in future studies on short- and/or long-term health effects. Overall, we suggest using D-STEM capabilities in increasing LUR studies that employ data of regulatory network monitoring stations.Highlights-First Land Use Regression using D-STEM, a recently introduced statistical software-Assess D-STEM in spatiotemporal modeling, mapping, and missing data imputation-Estimate high resolution (20×20 m) daily maps for exposure assessment in a megacity-Provide both short- and long-term exposure assessment for epidemiological studies

Download Full-text

Missing Data Imputation in Internet of Things Gateways

Information ◽

10.3390/info12100425 ◽

2021 ◽

Vol 12 (10) ◽

pp. 425

Author(s):

Cinthya M. França ◽

Rodrigo S. Couto ◽

Pedro B. Velloso

Keyword(s):

Neural Network ◽

Missing Data ◽

Internet Of Things ◽

Execution Time ◽

Regression Models ◽

Weather Data ◽

Data Imputation ◽

Missing Data Imputation ◽

Imputation Methods ◽

The Neural Network

In an Internet of Things (IoT) environment, sensors collect and send data to application servers through IoT gateways. However, these data may be missing values due to networking problems or sensor malfunction, which reduces applications’ reliability. This work proposes a mechanism to predict and impute missing data in IoT gateways to achieve greater autonomy at the network edge. These gateways typically have limited computing resources. Therefore, the missing data imputation methods must be simple and provide good results. Thus, this work presents two regression models based on neural networks to impute missing data in IoT gateways. In addition to the prediction quality, we analyzed both the execution time and the amount of memory used. We validated our models using six years of weather data from Rio de Janeiro, varying the missing data percentages. The results show that the neural network regression models perform better than the other imputation methods analyzed, based on the averages and repetition of previous values, for all missing data percentages. In addition, the neural network models present a short execution time and need less than 140 KiB of memory, which allows them to run on IoT gateways.

Download Full-text