0xideas / sequifier-config

This is the repository with the config templates for the sequifier package.

The simplest workflow is the following:

1. Run pip install sequifier
2. Clone this repository to your project name: git clone [email protected]:0xideas/sequifier-config.git YOUR PROJECT NAME
3. Create a folder data and copy your data to that folder
4. Ensure data quality, such as the removal of outliers and no NaN values in the relevant columns, and that there are two columns sequenceId and itemPosition, which indicate which sequence a row belongs to and which position in the sequence the row has, respectively
5. Adapt the fields data_path, selected_columns and target_columns in preprocess.yaml
6. Run sequifier preprocess from the project root
7. Adapt the fields ddconfig_path to configs/ddconfigs/DATASET-NAME.json, selected_columns, target_columns, target_column_types and criterion in train.yaml
8. Run sequifier train from the project root to train the model
9. Adapt the fields ddconfig_path, model_path, data_path, selected_columns, target_columns and target_column_types in infer.yaml
10. Run sequifier infer

Voila, you have extrapolated your categorical or real series using your custom transformer model!