Hasanali Faraji Sabokbar; Seyyed Hasan Motiee Langroodi; Hossein Nasiri
Abstract
Abstract
With the development of science and technology, a large amount of spatial and non-spatial data are stored on large databases. Analyzing these data for decision making necessitates the need for spatial data mining to discover knowledge. The use of satellite imagery, geo-statistical analysis, ...
Read More
Abstract
With the development of science and technology, a large amount of spatial and non-spatial data are stored on large databases. Analyzing these data for decision making necessitates the need for spatial data mining to discover knowledge. The use of satellite imagery, geo-statistical analysis, and all types of spatial data are useful and practical tools in studying land use change monitoring; but, what is important is the extraction of precise rules by integrating large amounts of data in order to provider knowledge about the area of interest. Rough Set Theory (RST) is one of the data mining techniques used in various ways in modeling uncertainty in data. Therefore, in this research, the RST knowledge discovery method is used to extract rules in combination with decision tree algorithm (DT) for satellite image classification and monitoring of land use changes. The results of the research indicate that according to the changes occurred during three periods of (1986-1998, 1998-2014 and 1986-2014), it can be seen that significant increasing and decreasing changes have occurred in the constructed lands and in the water bodies, while agricultural lands have not changed much. Of course, considering the base year (1986), it can be stated that the area of the agricultural lands under cultivation has witnessed a slight change compared to the base year which coincided with the imposed war, which means that the area under cultivation during the past three decades has been the same as that of the war period. This indicates that, the crisis is taking place in the agricultural sector. Also, in terms of methodology, given the overall accuracy and Kappa ratio, derived from the DT-RST combination model, RST can be considered to be a powerful tool in data mining, reducing the redundant data from databases and extracting rules for use in the DT method.