Investigation of urban climate and built environment relations by using machine learning

Climate change can cause a cascade of effects from the individual organisms to ecosystem-scale where in nature all species are elements of complex networks of interactions so every effect on every scale has great role. The properties of those networks are decisive on ecosystems all around the world so how they will be modified by climate change needs serious studies. Urban areas' vast population exert significant effects on the climate change even though they cover a small proportion of the Earth's surface, however, impacts of the urbanization on climate and ecosystems remain inadequately understood. In the meantime, urbanization continues to increase and in the 2030, 2/3 of the population is expected to be living in urban areas and this rate will increase in time. It is of importance to elaborate the studies investigating the relations between the urbanization and climate. In this respect, the use of information technologies with an extensive computational capacity is one of the cornerstones of the climate and urban studies. Machine learning algorithms is a branch of computer science that deals with automated recognition of patterns from data. The use of the machine learning algorithms can bring great advantages to both understanding and predicting the climate. The computational power with big data and ability to capture nonlinear behavior, learn as new data arrive, etc. make the machine learning favorable tool for understanding climate and developing urban planning. In this sense, purpose of this study is showing the advantages of machine learning algorithms by achieving working recurrent neural network algorithm to make climate predictions, and through this achievement, stating possible effects of machine learning on design and its contribution to understanding the climate.