predict new test data

[help116114]

Dear author:
I want to follow and cite your study. There is only training forecasting model script, such as weather with TimeXer.sh. However, after trained the forecasting model, we need the model to generate new forecasting data. Therefore, can you update a new script for this function?
I have attempted to configure the istraining==0 to generate test data. The test script is following:
export CUDA_VISIBLE_DEVICES=3

model_name="TimeXer"
des="Timexer-MS"

修改以下参数以适应你的需求

python3 -u run.py
--task_name long_term_forecast
--is_training 0
--root_path ./dataset/weather/
--data_path weather.csv
--model_id weather_96_96
--model "$model_name"
--data custom
--features M
--seq_len 96
--label_len 48
--pred_len 96
--e_layers 1
--factor 3
--enc_in 21
--dec_in 21
--c_out 21
--des "$des"
--d_model 128
--itr 1
And the test data has three .npy format files: pred.npy, true.npy, metric.npy. Then, I use numpy to transform npy to csv format.
But, the new forecasting data is not similar to original data in weather.csv.

So, can you provide us an instance of forecast-new.sh? Thank you very much.