This is the repository with the config templates for the sequifier package.
The simplest workflow is the following:
- Run
pip install sequifier
- Clone this repository to your project name:
git clone [email protected]:0xideas/sequifier-config.git YOUR PROJECT NAME
- Create a folder
data
and copy your data to that folder - Ensure data quality, such as the removal of outliers and no NaN values in the relevant columns, and that there are two columns
sequenceId
anditemPosition
, which indicate which sequence a row belongs to and which position in the sequence the row has, respectively - Adapt the fields
data_path
,selected_columns
andtarget_columns
inpreprocess.yaml
- Run
sequifier preprocess
from the project root - Adapt the fields
ddconfig_path
toconfigs/ddconfigs/DATASET-NAME.json
,selected_columns
,target_columns
,target_column_types
andcriterion
intrain.yaml
- Run
sequifier train
from the project root to train the model - Adapt the fields
ddconfig_path
,model_path
,data_path
,selected_columns
,target_columns
andtarget_column_types
in infer.yaml - Run
sequifier infer
Voila, you have extrapolated your categorical or real series using your custom transformer model!