Shodhganga Collection:
http://hdl.handle.net/10603/79677
2021-10-20T03:44:50ZSome testing problems for location and scale parameters using U statistics
http://hdl.handle.net/10603/331220
Title: Some testing problems for location and scale parameters using U statistics
Abstract: In Statistics, hypothesis testing is used for testing the equality of population parameters. Testing the equality of populations is one of the basic problems in Natural Sciences, Social Sciences and Formal Sciences. We developed some general classes of tests for testing the hypothesis associated with location/scale parameters of the two or more underlying populations in order to gain more efficiency in comparison to existing tests. An optimal choice of sub-sample size along with optimal choice of order statistics is obtained to maximize the efficacy. A general choice of optimal weights is worked out for testing the parameters without any restrictions on number of populations, sample size and peak of the restrictive alternatives. The application of the proposed classes of tests is demonstrated through existing real life data sets. Statistical power of the proposed classes of tests is computed using Monte Carlo simulation study.
newlineSome results on Bayesian inference using Zenga curve and related measures
http://hdl.handle.net/10603/306788
Title: Some results on Bayesian inference using Zenga curve and related measures
Abstract: The current research work relates to Bayesian inference related to a few inequality measures. Out of the many income inequality measures, the Lorenz curve (1905) and Bonferroni curve (1930) are the most popular because of its good statistical properties and straightforward economic interpretation. However, keeping in mind that the notions of poor and rich are relative to each other, Zenga (2007) proposed a new index of income inequality. The Zenga curve and its associated index introduced by Zenga are based on a comparison of the mean income of the poorer income earners with the mean income of the remaining richest part of the population. Motivated by this, we have considered more aspects on the Zenga measure as well as other income inequality measures using Bayesian inference approach. In addition to examining the connection between the Zenga curve and other existing inequality measures, we have obtained the Bayes estimators of Zenga curve as an alternate measure of inequality in context to some important income distributions. Also the Bayesian credible intervals have been obtained in the thesis for the income distributions.
newlineModelling techniques for unstructured data using machine learning and artificial intelligence approach
http://hdl.handle.net/10603/262031
Title: Modelling techniques for unstructured data using machine learning and artificial intelligence approach
Abstract: Intially, the concept of document models is discussed with respect to the Bernoulli approach, that is, basis is the presence or absence of primary blocks of the documents, namely tokens. The problem primarily deals with how an unstructured dataset consisting of text documents is converted to structured content with mathematical and statistical foundation.
newlineThe focus in another problem is on Multinomial document model. It is similar to the Bernoulli model, but the presence flag in the former is now replaced with the frequentist method which takes into account the number of times the tokens occur in the text.
newlineIn next problem, we move onto research of unsupervised topic modeling techniques to obtain latent topical structure across text documents and further fine-tuning with help of machine learning. Problem 4 deals with obtaining of key conversational drivers for textual data coupled with the attached sentiment and mood states. Such an approach helps in detecting the key drivers of conversation that is, whether it is having a positive impact or not or whether the text content has potential to become viral and much more.
newlineIn last problem, we focus on Mood State and Behavior Prediction model in Social Media through Unstructured Data Analysis. Behavior Dirichlet Probability Model (BDPM), which can capture the Behavior and Mood of user on Social Media.Modeling and analysis of survival data using frailty models
http://hdl.handle.net/10603/235159
Title: Modeling and analysis of survival data using frailty models
Abstract: None