A Error in data Processing

[GameHoo]

Thanks to the author for the paper and the code, this is really great work.

I found a fatal error in data processing.
When filling missing data, other coin's data is used.

I added a few lines of print code in globaldatamatrix.py to observe the missing and filled data.

The output is as follows

...
FCT 2015-10-05 00:30:00 2017-06-30 16:00:00
FCT 2015-10-05 00:30:00 2017-06-30 16:00:00
FCT 2015-10-05 00:30:00 2017-06-30 16:00:00
GNT 2017-02-18 04:00:00 2017-06-30 16:00:00
GNT 2017-02-18 04:00:00 2017-06-30 16:00:00
GNT 2017-02-18 04:00:00 2017-06-30 16:00:00
ZEC 2016-10-28 18:30:00 2017-06-30 16:00:00
ZEC 2016-10-28 18:30:00 2017-06-30 16:00:00
ZEC 2016-10-28 18:30:00 2017-06-30 16:00:00
coin_name: ZEC
close high low
2015-06-30 16:00:00 490.000000 2600.000000 2.048000
2015-06-30 16:30:00 490.000000 2600.000000 2.048000
2015-06-30 17:00:00 490.000000 2600.000000 2.048000
2015-06-30 17:30:00 490.000000 2600.000000 2.048000
2015-06-30 18:00:00 490.000000 2600.000000 2.048000
2015-06-30 18:30:00 490.000000 2600.000000 2.048000
2015-06-30 19:00:00 490.000000 2600.000000 2.048000
2015-06-30 19:30:00 490.000000 2600.000000 2.048000
2015-06-30 20:00:00 490.000000 2600.000000 2.048000
2015-06-30 20:30:00 490.000000 2600.000000 2.048000
2015-06-30 21:00:00 490.000000 2600.000000 2.048000
2015-06-30 21:30:00 490.000000 2600.000000 2.048000
2015-06-30 22:00:00 490.000000 2600.000000 2.048000
2015-06-30 22:30:00 490.000000 2600.000000 2.048000
2015-06-30 23:00:00 490.000000 2600.000000 2.048000
2015-06-30 23:30:00 490.000000 2600.000000 2.048000
2015-07-01 00:00:00 490.000000 2600.000000 2.048000
2015-07-01 00:30:00 490.000000 2600.000000 2.048000
2015-07-01 01:00:00 490.000000 2600.000000 2.048000
2015-07-01 01:30:00 490.000000 2600.000000 2.048000
2015-07-01 02:00:00 490.000000 2600.000000 2.048000
2015-07-01 02:30:00 490.000000 2600.000000 2.048000
2015-07-01 03:00:00 490.000000 2600.000000 2.048000
2015-07-01 03:30:00 490.000000 2600.000000 2.048000
2015-07-01 04:00:00 490.000000 2600.000000 2.048000
2015-07-01 04:30:00 490.000000 2600.000000 2.048000
2015-07-01 05:00:00 490.000000 2600.000000 2.048000
2015-07-01 05:30:00 490.000000 2600.000000 2.048000
2015-07-01 06:00:00 490.000000 2600.000000 2.048000
2015-07-01 06:30:00 490.000000 2600.000000 2.048000
... ... ... ...
2017-06-30 01:30:00 0.125222 0.126384 0.125208
2017-06-30 02:00:00 0.124460 0.125213 0.123112
2017-06-30 02:30:00 0.124190 0.124461 0.123001
2017-06-30 03:00:00 0.123411 0.124659 0.122930
2017-06-30 03:30:00 0.123202 0.125138 0.123150
2017-06-30 04:00:00 0.123593 0.123593 0.122863
2017-06-30 04:30:00 0.123131 0.123642 0.122634
2017-06-30 05:00:00 0.122255 0.123630 0.122255
2017-06-30 05:30:00 0.120116 0.122255 0.119000
2017-06-30 06:00:00 0.120899 0.120899 0.118810
2017-06-30 06:30:00 0.121580 0.121607 0.120000
2017-06-30 07:00:00 0.120638 0.121580 0.120000
2017-06-30 07:30:00 0.120860 0.121281 0.119570
2017-06-30 08:00:00 0.121236 0.121600 0.119730
2017-06-30 08:30:00 0.121999 0.123708 0.121200
2017-06-30 09:00:00 0.123400 0.124000 0.121900
2017-06-30 09:30:00 0.124126 0.124150 0.122851
2017-06-30 10:00:00 0.122200 0.124126 0.121930
2017-06-30 10:30:00 0.122500 0.123300 0.121930
2017-06-30 11:00:00 0.123500 0.124150 0.122500
2017-06-30 11:30:00 0.125060 0.125075 0.122820
2017-06-30 12:00:00 0.124100 0.125310 0.124000
2017-06-30 12:30:00 0.124000 0.124945 0.124000
2017-06-30 13:00:00 0.123446 0.123968 0.122834
2017-06-30 13:30:00 0.121110 0.123447 0.121110
2017-06-30 14:00:00 0.122062 0.122429 0.120874
2017-06-30 14:30:00 0.121581 0.121910 0.121510
2017-06-30 15:00:00 0.121200 0.121581 0.120370
2017-06-30 15:30:00 0.121790 0.121790 0.120938
2017-06-30 16:00:00 0.121000 0.122300 0.120938

It can be seen that ZEC had no data before 2016-10-28, but it was filled with a huge number, which is believed to have impact on the experiment.

[jiahaoli57]

Good point.