China
Africa
Explore chapters and articles related to this topic
Assimilation of Synchronous Data in Hydraulic Models
Published in Maurizio Mazzoleni, Improving Flood Prediction Assimilating Uncertain Crowdsourced Data into Hydrological and Hydraulic Models, 2017
In Figure 6.6, the Taylor diagrams (Taylor, 2001) for reach A and B during the six flood events are represented. Taylor diagrams are commonly used to graphically summarize how closely simulations fit observations. Such similarity is calculated by means of three statistic as root mean square difference (RMSD), correlation and standard deviation between observations and simulations. This means that, in Figure 6.6 the closest is DA method to the observations (black point) the better. Similar conclusions to the ones just mentioned can be summarized in Figure 6.6.
A hybrid approach of ANN and improved PSO for estimating soaked CBR of subgrade soils of heavy-haul railway corridor
Published in International Journal of Pavement Engineering, 2023
Abidhan Bardhan, Abdel Kareem Alzo'ubi, Sangeetha Palanivelu, Pouria Hamidian, Anasua GuhaRay, Gaurav Kumar, Markos Z. Tsoukalas, Panagiotis G. Asteris
On the other hand, the values of ‘sigma’ and ‘gamma’ (two hyperparameters of SVM) were chosen in the range of 0.001 – 0.1, and 0.1 – 200, respectively. The rbf kernel function was used to establish the best SVM model. Besides C-ANN and SVM, the most suitable structure of GMDH was obtained with = 10 in each layer. The best results were obtained when the was set to 12. The best GMDH model was chosen based on the results of the testing dataset. The outcomes of all constructed models based on the performance of various indicators, as well as visualisation of results through Taylor diagrams and scatter plots, are explained in the sub-sections below. It is pertinent to mention here that, the Taylor diagram is a 2-D mathematical illustration that is used to provide a concise assessment of a model's accuracy (Taylor 2001). It illustrates the relationships between the observations of actual and estimated models in terms of R, RMSE, and the ratio of standard deviations. In the Taylor diagram, a model is represented by a point. It should be noted that, for an ideal model, the position should correspond with the reference point (Ref).
Machine learning predictive approaches for hot crack mitigation in modified TIG welded AA7075 joints
Published in Materials and Manufacturing Processes, 2022
Dhilip A, Jayakrishnan Nampoothiri, Senthilkumar M, Kirubanandan N
Furthermore, the performances of selected models were compared with the help of the Taylor diagram and violin plot. The Taylor diagrams indicate how well the models fit together in terms of standard deviation and correlation coefficient. The performance of the considered models is analyzed by examining how predicted values are close to the observed values. The Taylor diagram affirms that the RFR model (plus symbol) is the best predictive model for both responses since it is very closer to the observed data (pink dot) with minimal standard deviation, least squares error and the strongest correlation as shown in Figs. 8 and 9. On the other hand, the MLR model (dark red dot) is away from the observed data for both responses and therefore, it can be nominated as the less accurate prediction model. According to the Taylor diagram, the performance of selected models could be ranked as RFR>ANN>MLR for both of the considered responses.
Comparative analysis of the performance of the GOFS, PSY4 and AMSEAS ocean model frameworks in the Virgin Islands and Puerto Rico coastal ocean
Published in Journal of Operational Oceanography, 2022
Sonaljit Mukherjee, Sennai Habtes, Paul Jobsis
The relative performance of a model output is best evaluated by the extent of the statistical deviation of each output from the observed data and the correlation of each output with the observed data. Taylor diagrams provide an effective means to visualise the statistical deviations and correlations of the model outputs. A Taylor diagram (Taylor 2001) is a polar coordinate representation of the following statistical measurements: root mean square deviation (RMSD) of the model output from the observed data, cross-correlation (R) of the model output and the observed data, and the standard deviation (σ) of the model output. The RMSD of the model output with respect to the observed data is calculated as where h and o are the model output and the observed dataset respectively and N is the sample size. The overline represents the mean.