Air Quality Prediction in Smart City's Information System
The introduction of new technology and computational power enables more data usages in a city. Such a city is called a smart city that records more data related to daily life activities and analyzes them to provide better services. Such data acquisition and analysis must be conducted quickly to support real-time information sharing and support other decision-making processes. Among such services, an information system is used to predict the air quality to ensure people's health in the city. The objective of this study is to compare various machine learning techniques (e.g., random forest, decision tree, neural network, naÃ¯ve Bayes, etc.) when predicting the air quality in a city. For the comparison, we perform the removal of records with empty values, data division into training and testing datasets, and application of the k-fold cross-validation method. Numerical experiments are performed using a given online dataset. The results show that the three best methods are random forest, Gradient Boosting, and k-nearest neighbors with precision, recall, and f1-score values more than 0.63.