In-Depth Analysis of Railway and Company Evolution of Yangtze River Delta with Deep Learning

Table 1

Comparison of images, minimum mean square error, and cosine similarity of three prediction results.

The name of the train station in the regional center, the name of the line to which it belongs, and the opening time		Hefei railway station in Yuliu railway (2013/12)	Huishan railway station in Shanghai-Nanjing railway (2010/07)	Xinggongtang station in Xuanhang railway (1994/07)
Data distribution before the completion of the railway
The true distribution of new data within three years after the completion of the railway
MMSE loss	Image
	MMSE value cosine similarity	0.000320737297485258	0.000017449833468964	0.000006898057400988
		0.130015477494133562	0.077197126502542367	0.099417792856082382
Cosine loss	Image
	MMSE value cosine similarity	1.5686035808463898	0.3253649305977068	0.056709321859160375
		0.586250486851695762	0.31263808319331287	0.697352200764660382
2D-DFT loss	Image
	MMSE value cosine similarity	0.29219522888094723	0.059963647176112536	0.02021538911975866
		0.148902076050230262	0.098727064405085867	0.079767883030792582
MMSE-2D-DFT loss	Image
	MMSE value cosine similarity	0.000303084697194027	0.000009756472944948	0.000003572309352900
		0.263564482924703362	0.111119785306437367	0.134587089183047082
Cosine-2D-DFT loss	Image
	MMSE value cosine similarity	4.620754636255468	0.7541079039032181	0.14423271414878933
		0.597224169228503626	0.33967178745360267	0.662771618719908282
MMSE- cosine loss	Image
	MMSE value cosine similarity	0.000364660292423330	0.000165248184589094	0.000053113823069145
		0.577345932340722066	0.338455365847816467	0.556173118171807028
MMSE- Cosine-2D-DFT loss	Image
	MMSE value cosine similarity	0.000366539845216779	0.000247506640524558	0.000071690391093354
		0.564076775017910262	0.341916216360121467	0.606259892930003782