Code for PhD Thesis

Thesis Title: Complex Networks in Manufacturing - Suitability and Interpretation

Author: Yamila Mariel Omar

Expected Defense Date: 8th July 2021

Institution: University of Luxembourg

This repository contains all the code necessary to reproduce my PhD Thesis. Find below a description of the files.

Directories

1. data: The raw data used for my thesis corresponds to the train_date.csv file in the Kaggle competition titled "Bosch Production Line Performance". Semi-processed data is available in this folder instead.
2. figures: Some of the figures in my Thesis are available in this folder.
3. flow_networks: all the code corresponding to the Chapter of "Flow Networks" is available in this directory.

Files description grouped by use

Data cleaning

1. from_timestamp_to_paths.py: code used to process the train_date.csv data from Kaggle (see here) to obtain manufacturing paths and manufacturing edges.
2. path_data_cleaning.py: code necessary to remove "noisy" data. It produces two files: clean manufacturing paths and clean edges.

Modules developed

1. graphfile.py: class used to open and write data from/to text files.
2. graph.py: class used to do complex network analysis.
3. betweenness_centrality.py: functions necessary to calculate the Betweenness centrality. Read docstrings for details.
4. clustering_coefficient: functions necessary to calculate the Clustering Coefficient. See docstrings for details.
5. depth_first_search.py: functions needed for the Depth First Search algorithm. Some functions to determine strongly connected components are also available here.
6. pagerank.py: functions needed to calculate the PageRank algorithm.

How to obtain the results in the Thesis manuscript:

1. traditional_topological_metrics.py will produce the results in Chapter 7, titled "Traditional Topological Metrics".
2. main_pagerank.py will produce the results in Chapter 8, titled "PageRank".

Profiling

1. main_for_entropy_code_profiling.py: To profile the code in the Entropy Chapter (Chapter 9) only the data from product family F2 was used. This snipped is the one profiled (otherwise, it would require HPC use).

Graphs and plots

1. plot_network.py: code used to generate the DOT file to plot the manufacturing network in the Thesis.