A BSc project to find software architectural concerns in Java source code through Machine Learning
The environment use Python 3.8.5

To start working in the environment, install required packages by running:
pip install -r requirements.txt

If unable to run the jupyter lab code, try removing the modin.pandas package and
use pandas instead.
When doing this also remove the ray package and the following 2 lines
ray.shutdown() and ray.init().
Modin transparently distributes the data and computation which means that we can
continue using the pandas API and only change the import line.
Uses a bit more extensive error checking, this shows up when running the notes.

The preprocess settings are defined in the config.yaml, where each setting
contains a list of lists.
The first item describes what to extract from the raw source
file, then the following strings is the parsing of the said extraction.

E.g. [['lib', 'tow', 'jk', 'scw','lc', 'stem'],['pac', 'tow', 'jk']]
First list, extracts the imports, tokenizes the words, removes java keywords,
separate composite words, lower case, then stemmize the words.
Second list, extracts packages, tokenizes them, remove java keywords.

All the settings are described down below.

settings = [
        ['extraction', 'parsing' ... ,'parsing],
        ...,
        ['extraction', 'parsing']
    ]

• Raw data: 'raw'
• Classes: 'c'
• Public methods: 'pm'
• Public variables/objects: 'pv'
• Import: 'lib'
• Packages: 'pac'
• Comments: 'com'
• Clear chosen settings from list: 'clear'

• Lower case: 'lc'
• Remove single characters: 'sc'
• Remove stop words: 'sw'
• Remove java keywords: 'jk'
• Remove numbers: 'nu'
• Separate compound words: 'scw'
• Stem words: 'stem'
• Tokenize words: 'tow'
• Clear chosen settings from list: 'clear'