irhete/predictive-monitoring-thesis

This repository contains scripts used in Irene Teinemaa's PhD thesis titled "Predictive and Prescriptive Monitoring of Business Process Outcomes"

Predictive monitoring of business processes

Original datasets:

• Github data: https://github.com/riivo/github-issue-lifetime-prediction
• Public event logs: https://data.4tu.nl/repository/collection:event_logs_real

Preprocessed datasets for predictive process monitoring:

• https://drive.google.com/file/d/1DDP7OKQhD8cno2tbSpLlIPZ-Mh5y-XUC/view?usp=sharing

For the preprocessing scripts, see https://github.com/irhete/predictive-monitoring-benchmark

Executing commands for different scripts can be found in the following Jupyter notebook: run_experiments.ipynb

Chapter 4: Benchmarking Existing Predictive Process Monitoring Techniques

1. Hyperparameter optimization
   • experiments_optimize_params.py
2. Training and evaluating final models
   • experiments.py
3. Execution times of final models
   • experiments_performance.py
4. Plots and tables
   • see https://github.com/irhete/predictive-monitoring-benchmark

Chapter 5: Predictive Monitoring with Structured and Unstructured Data

1. Hyperparameter optimization
   • experiments_optimize_params.py
   • experiments_optimize_params_with_unstructured_data.py
2. Training and evaluating final models
   • experiments.py
   • experiments_with_unstructured_data.py
3. Execution times of final models
   • experiments_performance.py
   • experiments_performance_with_unstructured_data.py
4. Plots and tables
   • generate_latex_tables.ipynb
   • plot_unstructured_results.R

Chapter 6: Temporal Stability in Predictive Process Monitoring

1. Hyperparameter optimization
   • experiments_optimize_params.py
   • experiments_optimize_params_with_unstructured_data.py
   • experiments_optimize_params_lstm.py
   • experiments_optimize_params_single_multirun.py
2. Training final models, calibrating, and writing predictions
   • experiments_write_predictions_stability.py
   • experiments_write_predictions_stability_unstructured.py
   • experiments_write_predictions_lstm.py
3. Evaluating prediction accuracy and temporal stability (RQ1)
   • evaluate_accuracy_stability.ipynb
4. Evaluating prediction accuracy and temporal stability of inter-run-optimized models (RQ2)
   • experiments_test_interrun_stability.py
   • experiments_test_interrun_stability_unstructured.py
5. Applying exponential smoothing (RQ3)
   • evaluate_accuracy_stability.ipynb
6. Plots and tables
   • generate_latex_tables.ipynb
   • plot_stability_results.R

Chapter 7: Alarm-Based Predictive Process Monitoring

1. Hyperparameter optimization
   • experiments_optimize_params.py
   • experiments_optimize_params_with_unstructured_data.py
2. Training final models and writing predictions
   • experiments_write_predictions_alarms.py
   • experiments_write_predictions_alarms_unstructured.py
3. Optimizing alarm thresholds
   • experiments_optimize_alarm_threshold.py
   • experiments_optimize_alarm_threshold_eff.py
   • experiments_optimize_alarm_threshold_ccom.py
4. Evaluating alarming thresholds
   • experiments_test_fixed_alarm_thresholds.py (RQ1 baselines)
   • experiments_test_optimal_alarm_threshold.py (RQ1)
   • experiments_test_optimal_alarm_threshold_eff.py (RQ2)
   • experiments_test_optimal_alarm_threshold_ccom.py (RQ3)
5. Plots and tables
   • generate_latex_tables.ipynb
   • plot_alarm_results.R