Finding duplicated test cases in MozTrap https://moztrap.mozilla.org

#Installation

Run install.sh (for Ubuntu)

##Updated:## Read tests/test_smoke.sh for usage

• Download test cases from Moztrap using download_sample.sh
• Manually mark the duplications as in input/ground-truth-217.csv
• Edit the config.py to add your new training data and perdict data files
• Run python finddup.py fit to generate a model (only use trainLocalJson)
• Run python finddup.py cross-val to do cross validation on feature and model parameters (only use trainLocalJson)
• Run python finddup.py perdict to create a model using trainLocalJson + groundtruth_filename and use that model to make perdictions on perdictLocalJson.
• If you need case ID and links, run add_case_id_to_output.py, remember to modify the output file path in the script.

If you don't want to call the MozTrap API everytime, use wget/curl to download the this url: https://moztrap.mozilla.org/api/v1/caseversion/?format=json&productversion=217 as json file. And set the localJson variable to its filename. Then change downloadCasevesions() call to loadLocalCaseversion(localJson).

• Define what is a duplication
• Create manully marked groundtruth data
• Tune the features and parameters
• Prettify the network graph

Open a pull request for review. Or submit ideas to slyu@mozilla.org