15858

14159

كارشناسي ارشد

آمار اقتصادي و اجتماعي

اصفهان : دانشگاه صنعتي اصفهان

1399

يازده،119ص.:مصور،جدول،نمودار

ساره گلي فروشاني

علي زينل همداني، افشين پرورده

1399/07/20

كتابنامه

آمار

رياضي

1399/07/23

2641269

Analysis of data on resilience indices in Isfahan against the crisis TOBA REZAEI MIRGHAED t rezaei@math iut ac ir September 2 2020 Master of Science Thesis in Farsi Department of Mathematical Sciences Isfahan University of Technology Isfahan 84156 8311 IranSupervisor Dr Sareh Goli Forooshani s goli@cc iut ac ir2010 MSC Primary 6207 Secondary 62P25keywords Data Mining Resilience Clustering Classification Neural Network Decision Tree K means Abstract Data mining techniques are used in data analysis of crisis resilience indicators in the city An integral partof data mining is the acquisition of knowledge about data KDD Also is a process which finds useful new valid novel previously unknown and ultimately understandable implicit patterns and models from largeand huge dataset Data mining have applications various like business Medicine sociology exercitation fraud detection et cetera which in this research attention to sociology and social dimension Data mining mainly depend on quality of data Raw data usually susceptive to incomplete data missing values inconsistent data outlier data and noisy data Thus it is important for data tobe processed before being mined The purpose of preprocessing is to transform the raw data to an adaptable form for subsequent analysis Alsopreprocessing data is an exigent and necessary step to enhancement data efficiency Preprocessing includeseveral important techniques like data cleaning data integration data transformation feature creation dataaggregation dimensionality reduction and feature subset selection This study display a detailed description ofdata preprocessing techniques which are used for data mining like imputation using KNN method for imputeto missing values with k 1 and Hamming distance and using the Grubbs test for detection outliers data Inthis research to working with categorical variables that have more than two levels using of One Hot Encoding OHE method and Feature Hashing method The Feature Selection method is used to identify which indices are the most important for amount resilience This research used of Minimum Redundancy Maximum Relevance MRMR Neighborhood ComponentAnalysis NCA and J3 criterion for feature selection and ranking indices

ساره گلي فروشاني

علي زينل همداني، افشين پرورده