Search results

Meanwhile, we take imbalanced data into consideration in the paper.
Nathalie [7] systematically investigated the challenges posed by imbalanced data in the classification problem and pointed out that data skew would affect various classifiers.
Intelligent Data Analysis, vol. 6 (2002), pp. 429-449
Garcia: Learning from imbalanced data, J.
Data Eng., vol. 21 (2009), pp. 1263-1284

In a World Bank Report, Dasgupta, Zhang, Wheeler and Huq (1996) use data collected from factories in Beijing and Tianjin to evaluate the likely cost effectiveness the levy system would have in abating industrial water pollution.
The statistics form the foundation for understanding the data that we collected from our primary research, that is, from interviews conducted with government officials, industry managers, consultants and academicians.
As for the regression analysis, due to the limited availability of time-series data for Beijing, we decided to create two separate regression models.
The Model Summary section describes the success of the overall model in predicting levels of pollution for Beijing, and from it we can see that the in-depth Beijing model was a reasonable approximation of our data.
In other words, if Beijing were to reduce its share of heavy industry or state-owned industry, we would expect a reduction in pollution.

Furthermore the modelling predictions were validated using experimental data.
During the 28days measuring period, the experimental temperature reduction were 0.59℃, 0.31℃, and 0.22℃ at 10m, 30m and 50m, respectively; while the simulation temperature reduction were 0.59℃, 0.31℃, and 0.22℃ at corresponding depths.
Furthermore, the temperature reduction of the ground became more and more obviously along with the operating of GSHP, it is necessary to study the temperature field underground with a long-time operating.
The model made a simulation on the heat extraction rate and temperature distribution under the ground, which was validated by experimental data.
(2) The temperature reduction of the ground became more and more obviously along with the operating.

Group Method of Data Handling.
Group method of data handling (GMDH) [17-18] is used in a wide variety of areas for data analysis, system modeling, optimization, and image recognition.
Checking the end-of-search condition is based on using data from the Mk.
The nature of the experimental data, namely the results of four deposition cycles under different process conditions, makes it fairly easy to divide the original data into subsets.
Another important advantage of this method is a significant reduction in the number of experimental data required to build a model, compared to other methods, due to repeated use of information for different combinations of input variables.

It exhibited two oxidation peaks at 0.360V and 0.638V and two reduction peaks at 0.431V & 0.103V.
Fig. 3b(b) is the voltammogram taken in inert atmosphere(Nitrogen) showing two oxidation peaks at 378 and 628mV and two reduction peaks at 427 and 94 mV.
Fig 3b(c) shows the voltammogram taken in oxidizing atmosphere where there is only one oxidation peak at 375mV and two reduction peaks at 415 and 146 mV.
Based on the Cottrell equation, the analysis of chronocoulometry data shows the charge-time dependence for linear diffusion control.
The intercalation of the polymer onto the clay layers has been confirmed by XRD data.

It was found that reduction of particle size, under certain conditions, can increase the injectability of cements.
Reduction of particle size also can improve injectability [3,4,7].
As it can be seen in Table 1, smaller particle size improves injectability of cement pastes, if the milled powder is not treated with heat or treated with heat at 500°C or 600°C temperature (this agrees with data found in literature [3,7]).
This agrees with particle size analysis – size reduction is more pronounced during first two hours of milling.
Conclusions As expected from data shown in literature heat treatment of α-TCP powders increased injectability, as did prolonged milling.

Although characterized by stability and low energy demand, BI devices are limited in their ability to adapt themselves to changing demands for structure response reduction.
The controlled system’s gain matrix between the structure displacement and the base acceleration vector may be inferred in the form (2) where P and Q denote the matrixes governing the distribution of the control action and the dependence on the response variables, as well as the gain of the isolation device (3) In order to select the most effective control action to be applied, the strategy may consist of optimizing the process through an approach accounting for recorded data from the structure under the ground motion till the moment of application of the control action and adopting a control force such that " t Î [-q,q] (4) The control force to be applied at the instant t may be obtained by instantaneously adding to the current control force expression, based on the Fourier transform of the data recorded up to the current instant the updated value of the error that should be suitably calculated.
Results are quite encouraging in terms of reduction of the structural response.
Actually, the numerical investigation shows a very good performance of the final control system with comparison to traditional devices, since besides the benefits related to the further reduction of the structural response that may be noticed in Figure 1, the procedure is optimized for requiring a very low energy amount in order to be effective.
The numerical investigation shows that the approach, obeying to an optimum problem, may lead to a significant further reduction of the structural response under dynamic excitation, while employing a very contained additional energy.

One example, for the first subject (exp1) data, is show in Table 2, where the "concise" decision table was constructed.
The theory was applied to handle imperfect data with uncertainty and vagueness by describing dependencies between attributes, evaluating the significance of attributes, and extracting causality patterns between attributes[7].
Similarly, we calculated the frequency of occurrence of all attributes in reducts for all the experimental data.
These are shown in Fig. 3, where we can see easily that the attributes 1, 4, 7 have relatively important positions in most experimental data.
For exp1 data, we induced six rules, there are 3 rules have relatively high strength () in their corresponding decision classes.

The unknown coefficients of empirical models are usually determined by referring to the relevant experimental data.
Empirical models are constructed by empirical or semiempirical fitting of test data or of measurements of carbonation depth of concrete structures in service life.
Comparison of prediction models Aforementioned three models are verified by comparing the calculated values with 79 groups accelerated carbonation data.
Values of carbonation depth predicted from the models are compared with the experimental data in Figure 1.
Comparison of predicted values and experimental data.

The double angle drill is effective in control of delamination in drilling of CFRP due to reduction of the thrust force.
Meanwhile, in drilling of A7075, the small wedge angle of the upper edges on the double angle drill is not effective in reduction of the thrust forces.
In another model, the cutting force was predicted with making the chip flow model using the orthogonal cutting data.
The following orthogonal cutting data were used for cutting of CFRP and A7075 with TiAlN coated tool: For carbon fiber reinforced plastic (CFRP): (2) For aluminum alloy (A7075): (3) where the units of angles, the shear stress and the uncut chip thickness are radian, Pa and m, respectively.
In drilling of aluminum alloy, A7075, a small wedge angle of the upper edges on the double angle drill is not effective in reduction of the thrust force.