Determination of liveability for turkish cities using GIS, citizen science and machine learning methods

In this study, an approach based on geographic information systems (GIS), citizen science (CitSci) and machine learning (ML) methods was developed in order to assess liveability in Turkish cities. Within the scope of the thesis, livability indices in national and international literature were investigated. The main categories influencing the liveability such as education, health, safety, transportation, environment, cultural and economic opportunities for cities in Turkey and their sub-criteria were determined; and two ML approaches was investigated for their performances to predict the liveability using spatial and non-spatial datasets. The multivariate linear regression and the artificial neural networks (ANN) methods were employed in the prediction. In order to obtain the metrics to assess the defined criteria, OSM (Open Street Map), which is an open source geospatial data source, and Turkish Statistical Institute (TÜİK) data were utilized. For the model training, the survey results on the liveability obtained in a Citizen Science initiative within the thesis were used as the dependent variable. The use of both the OSM data and the volunteer contributions in the study has demonstrated the usability of Citizen Science methods for urban liveability assessment. However, the OSM data quality and the sufficiency of the selected criteria require further investigations and are open for discussion. The study demonstrates one of the first examples in this field. It is anticipated that by increasing the data diversity and quality, as well as the amount of citizen contributions on the liveability assessments, the accuracy of the prediction results will increase and the methods can be tuned. The proposed approach can be useful for local governments to develop objective metrics, that can as well be monitored, for the improvement of quality of lives in cities; and can contribute to the urban sustainability and participatory planning goals.