Missing Data in Smart Cities: An Imputation Algorithm Based on Sine/Cosine Optimization Algorithm

With the emergence and proliferation of smart cities, which rely primarily on collected data, ensuring data quality has become very important. Missing data issues have emerged as a common problem in all smart city applications, particularly video surveillance. In video surveillance, missing data affects the accuracy and reliability of analysis and can also affect security by missing information about important locations. Therefore, the imputation of missing values is crucial to ensure the analyses' reliability and avoid biased results. This study proposed a novel data imputation algorithm to address the problem of missing data in smart city surveillance. The proposed imputation algorithm used only the features that are correlated with the target feature (the feature to be imputed). In the proposed algorithm, the Sine Cosine Optimisation (SCA) algorithm is used in the feature selection process. This algorithm aims to achieve higher accuracy by using more correlated features instead of using all available features. To compare the proposed algorithm with some state-of-the-art imputation algorithms, two sets of experiments are performed on real-world and video surveillance datasets with different rates of missing data. In each of the two sets of experiments, the Mean Square Error (MSE) is used to evaluate the performance of each imputation algorithm. Each experiment is repeated more than once, so the average of each algorithm's rank is calculated to avoid randomness. The results show a clear superiority of the proposed algorithm compared to the other imputation algorithms used with different missing rates. They demonstrate the proposed algorithm's effectiveness, robustness, and stability in dealing with data with missing values, especially in smart city surveillance.