Multivariate Anomaly Detection for Event Logs

Directory structure:

multivariate-anomaly-detection-for-event-logs
│   README.md
|   requirement.txt
|
|--- data: original dataset
│   │   bpi_2013.csv
|   |   bpi_2012.csv
|   |   small_log.csv
|   |   large_log.csv
|
|--- data_preprocessing
|   |   data_preparation.ipynb
|   |   data_exploration.ipynb
|   |   descriptive-statistics.ipynb
|
|--- utils
|   |   utils.py
|   |   models.py
|
|--- input: preprocessed data
|
|--- experiment
|   |   output
|   |   VAE.ipynb
|   |   AE.ipynb
|   |   LSTMAE.ipynb
|   

Reference

1. Install requirement

• Install pytorch: conda install pytorch torchvision -c soumith
• pip install -r requirements.txt

2. Run data_preparation.ipynb
3. Run VAE.ipynb or AE.ipynb or LSTMAE.ipynb

Event log Reconstruction

Directory structure:

event-log-reconstruction
│   README.md
│   requirement.txt
|
|--- data: original dataset
│   │   bpi_2013.csv
|   |   bpi_2012.csv
|   |   small_log.csv
|   |   large_log.csv
|
|--- data_preprocessing
|   |   induce_missing_data.py
|   │   preprocess_variables.py
|   |   real_log_preprocessing.sh
|
|--- utils
|   |   utils.py
|   |   models.py
|
|--- input: preprocessed data
|
|--- experiment
|   |   output
|   |   AE.ipynb
|   |   VAE.ipynb
|   |   LSTMAE.ipynb
|
|-- base_model
|   |   dummy_imputation.ipynb
|   |   statistical_description.ipynb
|

Reference

1. Install requirement

• Install pytorch: conda install pytorch torchvision -c soumith
• Install requirements: pip install -r requirements.txt

2. For preprocessing:

• cd data_preprocessing
• source real_log_preprocessing.sh

3. For training and evaluating:

• cd experiment
• Run AE.ipynb or VAE.ipynb or LSTMAE

4. For baseline models:

• cd base_model
• Run dummy_imputation.ipynb