This project has been executed via google colab.
-
Download the code as a zip file and extract the contents.
-
Create a new folder in google drive and upload the extracted contents into it.
-
Make the necessary changes mentioned in the code to get the result.
-
The datsasets used can be found in the
informer_model/Dataset-main/datasets
folder. Any new dataset to be used should be uploaded in the mentioned folder. -
All changes pertaining to prediction length, dataset etc can be done in
main_informer.py
script. -
The dataset can be changed in
main_informer.py
indata parser
underBattery
(The name of the csv file used should be mentioned under thedata
parameter. The target parameter can be varied under the parameterT
).
To train and evaluate the Informer model on a dataset the following command is used in the google colab notebook :
!python -u main_informer.py --model informer --data Battery --train_epochs 4 --attn prob --freq s --features S
The detailed descriptions about the arguments are as following:
Parameter name | Description of parameter |
---|---|
model | The model used in the experiment. This can be set to informer , informerstack |
data | The dataset name |
train_epochs | The number of epochs the code has to run for (defaults to 6) |
attn | Attention used in encoder (defaults to prob ). This can be set to prob (informer), full (transformer) |
freq | Freq for time features encoding (defaults to h ). This can be set to s ,t ,h ,d ,b ,w ,m (s:secondly, t:minutely, h:hourly, d:daily, b:business days, w:weekly, m:monthly |
features | The forecasting task (defaults to M ). This can be set to M ,S ,MS (M : multivariate predict multivariate, S : univariate predict univariate, MS : multivariate predict univariate) |
On running the main code a results
folder will be created where the prediction output is stored.
We appreciate the following github repos a lot for their valuable code base or datasets:
https://github.com/zhouhaoyi/Informer2020
Diao, W., Saxena, S., Pecht, M. Accelerated Cycle Life Testing and Capacity Degradation Modeling of LiCoO2 -graphite Cells. J. Power Sources 2019, 435, 226830.