Open AccessModel to Predict Wartime Equipment Waste Based on Multiple Regression Analysis
Author(s) -
Ziqi Li,
Guiming Chen,
Qiaoyang Li
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1802/4/042044
Subject(s) - polynomial regression , ordinary least squares , regression analysis , linear regression , proper linear model , statistics , regression , multivariate statistics , residual , correlation coefficient , computer science , engineering , mathematics , algorithm
Predicting wartime equipment waste has not only been a research topic at home and abroad but is also difficult. Traditional research methods emphasize the trend of wartime equipment waste but ignore the effects of influencing variables. This paper discusses the influencing factors of wartime equipment waste and proposes a model to predict wartime equipment waste based on multiple regression analysis. An ordinary least-squares approach is used to estimate the parameters of the model. The significance of the model and regression coefficient is evaluated by the complex correlation coefficient and the t-test, respectively. Ways to optimize and improve the model in future studies are discussed. Expressions to calculate the prediction value and prediction interval are given. This paper establishes three main factors that can be quantified and have a certain sample size: combat time, failure rate, and military input. First, a conventional multiple linear regression model is established. Based on the results of the significance test, the model is iteratively analyzed and optimized. Second, a stepwise regression method is used to screen the independent variables, and a weighted least- square estimation is used to evaluate the influence of changes in the automatic variables on the residual. Finally, a multivariate nonlinear regression approach using a weighted ternary quadratic polynomial model for statistical data is discussed and established. Results from testing indicate that the proposed model has good significance, feasibility, and practical application.