Though research on poverty is numerous, it is in recent times that data scientists have taken interest in understanding the phenomena using various nonconventional methods. The absence of large-scale data on same households at different points of time, has deprived researchers of the deeper analysis of household dynamics in general and poverty in specific. IHDS database provided a unique opportunity to fill this gap. One of the earlier studies using the same database, while analyzing the characteristics impacting escaping and falling into poverty considers the same set of attributes which explain both the phenomena. This work makes following contributions: 1. It has been assumed that different attributes explain escaping and falling into poverty. 2. It uses a machine learning approach that identifies the respective strength of each explaining attribute more accurately. 3. The method classifies escaping and falling into three and two groups respectively, that suggests the vulnerability within the groups and 4. Similarities and differences in results from the previous study reinforce the existing established characteristics as well as provides new nuances to ponder about. Overall, this research is a definitive contribution on method, analysis and findings.
S. Narendranath, Khare, S., Gupta, D., and Jyotishi, A., “Characteristics of ‘Escaping’ and ‘Falling into’ Poverty in India: An Analysis of IHDS Panel Data using machine learning approach”, in International Conference on Advances in Computing, Communications and Informatics (ICACCI), International Conference on Advances in Computing, Communications and Informatics (ICACCI)), PES, Bengaluru, 2018.