dodgy719 / decisions_with_ml

Decisions with Machine learning

Machine learning models are used everywhere these days and it has become very easy to interpret some of these machine learning models. We will be looking at a few specific tools which help us interpret these models and better understand the results these models provide. Here are those libraries

1. InterpretML
2. DiCE ML
3. SHAP

There are a few files here which talk about different things.

• preprocessing.ipynb talks about preprocessing the data and performing steps needed to build a machine learning model. Not the most important for now
• Decision Trees Interpretation.ipynb talks about building decision trees, a machine learning model and how to interpret them
• Counterfactuals.ipynb talks about generating what-if scenarios based on the current data and identifying what it would take for a point to get classified in the other class

decisions_with_ml's Issues

Update 'master' branch name to 'main'

Reasoning and execution:
https://stevenmortimer.com/5-steps-to-change-github-default-branch-from-master-to-main/

Suggestions for DTI.ipynb

https://github.com/dodgy719/Decisions_with_ML/blob/master/Decision%20Trees%20Interpretation.ipynb

Add # notes in script

• In 1: Needed to execute pip install dice-ml (ModuleNotFoundError: No module named 'dice_ml')
• In 1: Needed to execute pip install pydotplus (ModuleNotFoundError: No module named 'pydotplus)
• In 12: Needed to execute brew install graphviz (FileNotFoundError: [Errno 2] No such file or directory: 'dot')
• In 14: Needed to execute pip install interpret_text (ModuleNotFoundError: No module named 'interpret')
• In 17: Needed to execute pip install plotly (ModuleNotFoundError: No module named 'plotly')
• In 17: Needed to execute pip install dash-cytoscape (ModuleNotFoundError: No module named 'dash_cytoscape')
• In 20: Exception: Model type not yet supported by TreeExplainer: <class 'sklearn.tree._classes.DecisionTreeClassifier'>
  *unsure how to interpret and address