An alternative model for the analysis of detecting electronic industries earnings management using stepwise regression, random forest, and decision tree

Abstract

This study attempts to diagnose the detecting electronic industries' earnings management by integrating suitable soft computing methods. Accounting earnings information is a very crucial element for corporate stakeholders to determine their stock prices and evaluate their supervision and management authority's performance, while it is also essential information for measuring corporate value. Hence, whether an enterprise can faithfully express its true economic meaning over its financial statements and how the management handles its earnings have turned out to be a popular issue widely discussed by researchers. Detecting public companies' earnings management is an important and challenging issue that has served as the impetus in many academic studies over the last few decades. Data mining technique and machine learning methods have also been commonly applied by accounting and financial personnel to other fields of studies. The study used the stepwise regression and random forest techniques to screen the variables in the first place, followed by adopting three kinds of decision trees including Chi-squared automatic interaction detector, classification and regression trees and C5.0 to establish a model and find out if the tested enterprise had extreme earnings manipulation. The results show that the proposed hybrid approach (RF+C5.0) has the optimal classification rate (the accuracy rate is 91.24 %) and the lowest occurrence of Type I error and Type II error. Also, as discovered from the rules set of the Communicated by V. Loia. final additional testing, an enterprise's operating cash flow, times interest earned ratio and previous period's discretionary accruals play a decisive role in affecting its extreme earnings management.

Higher education as a factor of socio-economic performance and development

“…The procedure is run with the use of a sequence of F-test, t-tests, adjusted R-square, Bayesian information criterion, Mallow's Cp or false discovery rate (see e.g. Chen and Howard, 2016). Backward step regression used in our paper started with the initial model that contains all the independent variables.…”

Modified hybrid forecast model considering chaotic residual errors for dam deformation

“…Inspired by the application of shuffled frog leaping algorithm (SFLA) in complex nonlinearity and optimization problems in management, economic optimal allocation, medicine, power, and many other fields, modified hybrid forecast model based on SFLA is proposed. First, according to the deformation prototype observation data, statistical models are established with common methods, for example, stepwise regression and back propagation (BP) neural network . Elastic moduli of dam body and foundation are inverted with the finite element analysis software ANSYS, accordingly, hybrid models are established.…”

Performance Analysis of Statistical and Supervised Learning Techniques in Stock Data Mining

“…C4.5 is a decision tree-based technique that employs a top-down recursive divide and conquer approach for data classification [43,44]. Random forest is an ensemble-based classifier that can be used for enormous and multifaceted databases for exploration, classification, and prediction [45,46]. SVM is one of the discriminative classifiers in which classification is based on the decision planes (multidimensional or hyperplanes) and their boundaries, and is effectively used for both classification and regression.…”