Search results

Electricity demand forecasting has always been a key issue, and inaccurate forecasts may mislead policymakers. To accurately predict China’s electricity demand up to 2030, this paper aims to establish a cross-validation-based linear model selection system, which can consider many factors to avoid missing useful information and select the best model according to estimated out-of-sample forecast performances.

With the nine identified influencing factors of electricity demand, this system first determines the parameters in four alternative fitting procedures, where for each procedure a lot of cross-validation is performed and the most frequently selected value is determined. Then, through comparing the out-of-sample performances of the traditional multiple linear regression and the four selected alternative fitting procedures, the best model is selected in view of forecast accuracy and stability and used for forecasting under four scenarios. Besides the baseline scenario, this paper investigates lower and higher economic growth and higher consumption share.

The results show the following: China will consume 7,120.49 TWh, 9,080.38 TWh and 11,649.73 TWh of electricity in 2020, 2025 and 2030, respectively; there is hardly any possibility of decoupling between economic development level and electricity demand; and shifting China from an investment-driven economy to a consumption-driven economy is greatly beneficial to save electricity.

Following insights are obtained: reasonable infrastructure construction plans should be made for increasing electricity demand; increasing electricity demand further challenges China’s greenhouse gas reduction target; and the fact of increasing electricity demand should be taken into account for China’s prompting electrification policies.

Based on the strategy and new institutional economic literature, this study aims to explore how different levels of supplier concentration (SC) will be characterized by differences in switching cost and coordinated adaptation in an ecosystem, thereby shaping its research and development (R&D) intensity, innovation performance and innovation efficiency.

This study adopted a set of panel data of Chinese listed firms in the Growth Enterprise Board and their top five suppliers from 2012 to 2016. A Tobit model is used to test the hypotheses.

The study finds that SC has an inverted U-shape effect on R&D intensity. This finding implies that firms are more likely to invest in R&D when SC is intermediate level. While it has a U-shape relationship between SC and innovation output, both lower SC and higher SC are more efficient in innovation because of their advantage in low switching cost and better coordinative adaptability, respectively.

The study complements the innovation ecosystem literature by using SC to represent the structure of the interdependence between firms and suppliers in an ecosystem, then examining the correlation between SC and firms’ innovation investment and output, respectively. Second, combining strategy and new institutional economic literature, the non-linear effects of SC on firms’ innovation are found.

Unsupervised domain adaptation object detection not only mitigates model terrible performance resulting from domain gap, but also has the ability to apply knowledge trained on a definite domain to a distinct domain. However, aligning the whole feature may confuse the object and background information, making it challenging to extract discriminative features. This paper aims to propose an improved approach which is called intrinsic feature extraction domain adaptation (IFEDA) to extract discriminative features effectively.

IFEDA consists of the intrinsic feature extraction (IFE) module and object consistency constraint (OCC). The IFE module, designed on the instance level, mainly solves the issue of the difficult extraction of discriminative object features. Specifically, the discriminative region of the objects can be paid more attention to. Meanwhile, the OCC is deployed to determine whether category prediction in the target domain brings into correspondence with it in the source domain.

Experimental results demonstrate the validity of our approach and achieve good outcomes on challenging data sets.

Limitations to this research are that only one target domain is applied, and it may change the ability of model generalization when the problem of insufficient data sets or unseen domain appeared.

This paper solves the issue of critical information defects by tackling the difficulty of extracting discriminative features. And the categories in both domains are compelled to be consistent for better object detection.