A user-friendly interface for removal of outliers from the yesilirmak river water quality data

Water plays a crucial role in the life of living organisms. Water resources have been seriously threatened in recent years due to rapid population growth, increased industrial activities, and unsustainable agricultural practices. On the other hand, significant measures have been taken to protect water resources and monitor pollution. In recent years, online stations and measurement methods have been introduced for water pollution monitoring. This allows for more frequent monitoring and facilitates the collection of a large amount of data. Furthermore, advancements in data mining and computer technologies have enabled the development of predictive models for various parameters, replacing laborious and costly experimental methods. However, data quality is important in models developed using real data, necessitating proper data preprocessing. This thesis focuses on removing outlier values in the dataset using the data collected from an online monitoring station on the Yeşilırmak River between 2007 and 2009. Evaluations were performed for dissolved oxygen (luminescence dissolved oxygen, LDO), temperature, pH, conductivity, total organic carbon (TOC), nitrate nitrogen (NO3-N), and ammonium nitrogen (NH4-N) based on the data obtained from these stations established under a TÜBİTAK project. Five outlier data removal methods were utilised: median, mean, Grubbs, Gesd, and interquartile range. Additionally, a graphical user interface was developed in the MATLAB environment. As a result, users can introduce any dataset to the software and remove outlier values using different methods in a few steps, preparing the data for modelling studies. It was observed that the median algorithm removed more data points among the outlier data-removing methods.