Subpop-miner

This repository contains the source code to mining algorithm that identifies subpopulations where outliers are defined differently than in the rest of the population and might need adjustments in their protection.

Note: The user interface is in development and may contain bugs, incomplete form, and other interface affordances.

How to setup your environment?

Requirements

• Python 3.9
• Qt 6.1.2

Setup

Use the requirements.txt file with pip to install the necessary packages.

pip install -r requirements.txt

How to run the application?

Run the application

python main.py

The user interface

The user interface provides 5 step wizard to guide the user through the process.

Wizard step 1: Loading the dataset

---

Load data: The user can load the data from a CSV file. The CSV file must contain a header row.

Wizard step 2: Selecting the attributes to be used in the analysis

---

Select relevant columns: The user can select the columns that are relevant to the data protection project.

Wizard step 3: Indicating the attribute types and the variable subject extreme value protection

---

Indicate data types: The user needs to indicate the data types of the selected columns. The data types are considered in two levels: the first level is the general data type (numeric, categorical), and the second level indicates if the variable is dependent or independent. The dependent variable is a numerical variable outlier of which must be protected. The independent variables are categorical and continuous variables that define subpopulations.

The variable subject to protection is indicated by the user by selecting the target radio button.

Wizard step 4: Setting the parameters for the mining algorithm

---