search

dbiir / TS-Benchmark

时序基准评测工具
GNU General Public License v3.0
18 stars 10 forks source link

Can't load the provided data #1

Open AKheli opened 3 years ago

AKheli commented 3 years ago

I am trying to run the generator script, the code referees to the column_23_3072_3072 that exists no where in the dataset.

date=np.loadtxt('./column_23_3072_3072.txt',delimiter=',')

What format should the input file have?

Thanks!

rainmaple commented 3 years ago

Data generator should run in mode 0 first in order to generate data, which database you prepare to test?

AKheli commented 3 years ago

Thanks for your response!

I am not sure what you mean by mode 0. My understanding is that we first need to train the GAN model, generate fragments, then use the random_walk code to generate long time series.

When trying to run the code test_dc,, the input file seems to have to be in a 3072x3072 format: 

date = np.loadtxt('./column_23_3072_3072.txt', delimiter=',')
lis = []
for i in range(3072):
    lis.append(date[i].reshape((3, 32, 32))/10)

I am using your dataset GoldWindSensor for example. Which does not have this format. How could I create the fragments? That is, how could I convert the dataset data into the 3072x3072 format?