The execution steps of our framework is explained below.
- Build global models
- Generate the neighbourhood using instance_generation function in defect_class.py
- Generate the rules using Association Rule Mining. Here, the generated neighbourhood of a particular instance is considered as the input the association rule miner.
- Option 1: The most closest open source package to Magnum Opus is, Opus Miner https://cran.r-project.org/web/packages/opusminer/index.html
- Option 2: Use BigML https://bigml.com/dashboard/sources
The steps of generating rules using BigML
Step 1: Click on the dataset with the neighbourhood in the source file
Step 2: Click on configure dataset
Step 3: Click on configure dataset. After deselecting unwanted fields, click on “Create Dataset”
Step 4: Click on Association under un-supervised category to build the model. Set the search category (confidence, coverage, lift and leverage). Specify the consequent as the target column.
Step 5: Click on "Association" under unsupervised category to generate the model
Step 6: Output rules and import those rules to csv.
- Categorize generated rules to four different types; which provide four types of guidance using
generate_excels()
function in excel_defect.py file.