Application Of Genetic Algorithm Classification Approach to Study Urban Streets Morphology at Neighborhood Scale

Abstract

Today’s cities worldwide are facing several new challenges with the spread of advanced digitalization and information technologies. As science and innovation are going digital, urban planning is highly concerned and should follow up with this global numerical transition. Urban planners should make use of the potential of new technologies to develop better and smarter urban forms responding to the new challenges and issues. The study uses artificial intelligence techniques based on genetic algorithms and supported by statistical data upon 20 indicators applied on 450 street segments in different urban fabrics in Fez city aiming to classify and simulate urban streets morphology. Machine learning can have the power of solving complex issues that humans alone cannot. The results using the potential of Machine Learning techniques can be a framework for decision makers to help them thinking about an intelligent planning process matching today challenges while taking advantages of new technologies. maps generates the integration values of each district based on number of segments’, their length, and intersections with other lines. Combining these data, axial maps reveal integration level and intelligibility with other spatial features. The city segments are shown as axial lines sorted from the shortest to the longest assigning a range of colours from red to blue, where red is the most integrated segment and blue is the most isolated one. Based on the data generated from these maps, the comparison of the street’s length and integration level of the three case studies have been done using Spyder(python3.5) software27, based on a test after 1000 iterations. The result of this