Sequential Pattern Mining with Complex Structure

Evaluation of an approximate approach on synthetic and medical data.

Main Programs:

The code needs Python 3.0 in order to be run.

• main.py
  • It manages the three algorithms to assess the Sequential Pattern Mining problem.
  • It contains an helper that provide all the information to run the code.
• generator/dataset_generator.py
  • It contains a simple code to generate a random dataset of sequences of complex structure.
  • There is the gen.sh to generate automatically a set of datasets throught this script

Supporting code

- print_graphs.py -> Support program to generate graphics on results
- comparator.py   -> Support program to compare .db outputs

Project Structure

The directory is structured as follow:

• generator:
  • It contains the script used to generate the synthetic datasets.
• datasets:
  • It contains the input datasets, generated or not by the dataset_generator
  • if generated, the name contains the information about the setting used
• experiments:
  • It contains the csv output data of the main program
  • Each row contains specific information about a single algorithm execution
• results:
  • It contains the .db output data of the main program
  • Each file contains the frequent sequences the main program found
  • The name define the algorithm used and information to recover the input data
• data_printers:
  • It contains some helper scripts used to generate the graphs.

Synthetic Medical Dataset

https://synthetichealth.github.io/synthea/