University of Twente Student Theses


Application of spatial data mining techniques for the study of infant mortality in India

Reddy, Mandadi Santhosh (2020) Application of spatial data mining techniques for the study of infant mortality in India.

Full text not available from this repository.

Full Text Status:Access to this publication is restricted
Abstract:Data Mining is a branch of data science that helps in better understanding the behavior of the data, which helps in determining solutions for real-world problems. When we deal with spatial data, this analysis is called Spatial Data Mining. This study involves techniques, namely Spatial Clustering and Predictive Modelling. The prime objective of the study is to identify the Spatial and Spatio-temporal hotspots of Infant Deaths in India. The same is achieved by collecting the Infant Mortality data of India from the year 2009-17 at the district level. Quarterly-wise data is used in identifying the Spatio-temporal hotspots. Spatial Scan Statistics method is carried out in the process of spatial clustering to identify the infant death hotspots. It is observed that spatial and spatio-temporal hotspot of infant deaths are purely random. Central Part of India has shown the first most likely cluster with a Log-Likelihood Ratio (LLR) of 51838, and Relative Risk of 2.48 is identified. Covariates Maximum Temperature, Minimum Temperature, Relative Humidity, and Solar Radiation are considered as independent variables for the regression modeling. Considering Infant Deaths as dependent variable, the prediction model is developed using the Random Forest regression model and Deep Learning technique using Artificial Neural Networks. It is observed that the Random Forest Regression has an accuracy of 0.66, and Neural Network has 0.71. So, for this data neural network has outperformed.
Item Type:Essay (Master)
Faculty:ITC: Faculty of Geo-information Science and Earth Observation
Programme:Geoinformation Science and Earth Observation MSc (75014)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page