A Method to Improve Parameter Estimation Success

Abstract

This paper introduces a method for improving parameter estimation in statistical models. Parameter estimation is a popular area of study in statistics, and recent years have seen the introduction of new distributions with more parameters to enhance modelling success. While finding a suitable model for a dataset is crucial, accurately estimating parameters is equally important. In some cases, classical parameter estimation methods fail to provide a closed form of estimation for parameters. As a result, researchers commonly resort to numerical methods and software programs for parameter estimation in models. The success rates of models have gained significance with the rising popularity of novel techniques like machine learning algorithms and artificial neural networks. Robust and reliable models are built on the strong theoretical foundations of statistical distributions. Specific distributions are used in various research fields to model datasets, and the assumptions associated with these distributions provide valuable insights into observations. Additionally, parameter estimation results sometimes lead researchers to direct conclusions. This paper presents an improvement method that relies on the estimation of parameters from other statistical distributions. This novel approach aims to make parameter estimation easier and more successful in certain situations. In the applications in this paper, the proposed methodology improves the success rate by up to 10% which provides an additional 6% success in the models.