Task:
Creating a dashboard to help an analyst to discover which features are reasons for misclassification. Highlighting possible sources of bias that results in bad quality decisions by the Machine Learning model.
Directory Structure
people-analytics-coursework
├── requirements.txt
├── Group7SMM635.pdf
├── datasets
│ └── cSVM.json
├── script
│ ├── 01_visualization.ipynb
│ └── output
│ └── dashboard_tabs.html
│ └── correlation_classified.jpeg
│ └── correlation_misclassified.jpeg
│ └── confusion_matrix.jpeg
│
└── key_findings.pdf
└── README.md
datasets
cSVM.json -> Dataset as provided
script
01_visualization.ipynb -> Python code for creating the dashboard
output
dashboard_tabs.html -> Final dashboard with two tabs (Continuous and Categorical)
correlation_classified.jpeg -> Correlation matrix between all continuous variables in case of correctly classifed individuals
correlation_misclassified.jpeg -> Correlation matrix between all continuous variables in case of incorrectly classifed individuals
confusion_matrix.jpeg -> Confusion matrix for True vs Predicted, Above and Below threshold
key_findings.pdf -> Executive summary illustrating the key insights emerging from the dashboard