China
Africa
Explore chapters and articles related to this topic
Forecasting
Published in Susmita Bandyopadhyay, Production and Operations Analysis, 2019
Causal model expresses the cause-and-effect relationship between dependent and independent variable. In case of Linear Regression, as described in Section 4.4.2.1.1, the variable “time” was considered as independent variable on which some other variable depended. In case of Causal model, similar concept is applicable except the fact that both the dependent and independent variables will have separate set of data at different time periods. For example, consider the relationship between crop yield and rainfall. Here, rainfall can be taken to be an independent variable and the crop yield can be treated as dependent variable since the crop yield is dependent on the amount of rainfall. Consider the following numerical example on the cause-and-effect (or causal) relationship between advertisement expenditure and sales of a company as shown in Table 4.15.
Forecasting Methods
Published in John G. Wensveen, Air Transportation, 2018
In general, a causal model is constructed by finding variables that explain, statistically, the changes in the variable to be forecast. Such variables must have the following characteristics: (1) they must be related statistically to the dependent variable, (2) data on them must be available, and (3) there must be some way of forecasting them, or their relationship to the dependent variable must be lagged (must follow the dependent variable by several months).
Forecasting Methods
Published in John G. Wensveen, Air Transportation, 2016
What is meant by a causal, or model, forecast? Define dependent and independent variables and correlation. What are the three characteristics that variables must have to be used in building a model? What are some of the limitations of causal models?
A causal inference method for canal safety anomaly detection based on structural causal model and GBDT
Published in LHB, 2023
Hairui Li, Xuemei Liu, Xianfeng Huai, Xiaolu Chen
Although both tasks are based on statistical modelling methods for model learning, the learning process differs between them, making the parameter estimation results of the constructed statistical models different. Structural causal modelling is a modelling approach used to accomplish the task of causal inference. Structural causal models include causal graphs and probability distribution expressions used to describe the causal relationships between variables. There are three types of variables – the explanatory variable x, output variable y, and unobserved variable u – where x and u are used to describe the independent variables in a causal relationship, and y is used to describe the dependent variable in that relationship. When the observed variable x is not sufficient to explain the generation of y, u is often introduced as a complement to the cause. A change in the value of the independent variable directly leads to a change in the output variable, and the causal relationship between them is often described using a conditional probability distribution, i.e. . The process of constructing a structural causal model has two steps: the construction of a causal graph and model training.
Two-stage approach to causality analysis-based quality problem solving for discrete manufacturing systems
Published in Journal of Engineering Design, 2023
Haonan Wang, Yuming Xu, Tao Peng, Reuben Seyram Komla Agbozo, Kaizhou Xu, Weipeng Liu, Renzhong Tang
Causality analysis theory was originally proposed by Fisher (1970) and Granger (1969), further developed by Judea Pearl (2000), and has been constantly maturing over the last two decades. Generally, causality analysis has two parts, causal discovery and causal inference. The former, also known as causal structure learning, is used to learn the causal relationship between variables, where algorithms can be classified as constraint-based, score-based, and hybrid methods (Zhou and Chen 2022; Glymour, Zhang, and Spirtes 2019). Causal inference, also refers to causal effect estimation, uses the observed data to estimate the causal effect of one variable on the other, where Structure causal model and Rubin causal model are typically used (Yao et al. 2021). The estimated causal effect can eliminate the bias caused by confounding factors (Li, Ding, and Mealli 2023; Moraffah et al. 2021), providing a credible causal relationship to support solution creation. However, in manufacturing systems, there usually exist confounding factors between process parameters and product quality. This leads to a significantly low efficiency in on-site quality problem solving. By incorporating causality analysis, the bias brought by confounding factors could be eliminated (Mooij et al. 2016). The preceding discussions indicate that incorporating causality analysis is crucial in creating solutions (Li and Shi 2007).
Model evaluation in human factors and ergonomics (HFE) sciences; case of trust in automation
Published in Theoretical Issues in Ergonomics Science, 2023
Mehdi Poornikoo, Kjell Ivar Øvergård
The evaluation of TiA models was conducted based on the proposed criteria to assess their adherence to each criterion. Prior to discussing the evaluation results, it is essential to examine the relationships between the criteria. As illustrated in Table 3, there exists a positive correlation between the testability and predictive power of the models. This is because in order to measure the predictive power, the model’s assumptions must be measurable and testable. Testability is also a meaningless idea without the model generating some predictions to be tested. Conversely, explanatory power and predictive power appear to be inversely correlated. This can be understood from a perspective of modelling functionality and the trade-off between the explanation and prediction (Watts et al. 2018; Hofman, Sharma, and Watts 2017; Yarkoni and Westfall 2017). Conceptual causal models that aim to encompass a wide range of instances by incorporating ample causal factors may have limited predictive capabilities. On the other hand, predictive models (e.g. regression, time-series) may achieve higher accuracy by narrowing down the causal elements, resulting in less generalizable outcomes (i.e. reduced explanatory power).