Graphical User Interface

Set up a custom dataset

1. Click on the Datasets navigation button in the left menu.

2. Click on the Add custom dataset button at the top of the page.

3. In the Name section, enter the name of your dataset. It should be unique.

4. Choose the type of the dataset in section Type.

   • Custom code – it allows to create your own custom dataset.

   • CSV file (last column will be treated as class column) - it allows to use a file with your own dataset. If you chose this option, you should set a path to the file on your computer in the Path input.

   • KEEL files (should have KEEL datasets structure) - it allows to use files with KEEL datasets structure. If you chose this option, you should set a directory to the KEEL files on your computer in the Path input.

5. Click on the Save button in the lower right corner of the page.

Note

Selecting at least one dataset is required.

Set up a custom classifier

1. Click on the Classifiers navigation button in the left menu.

2. Click on the Add custom classifier button at the top of the page.

3. In the Name section, enter the name of your classifier. It should be unique.

4. Choose the type of the classifier in section Type.

   • Batch learning – represents the training of machine learning models in a batch manner. The system is not capable of learning incrementally.

   • Incremental learning - model learns and enhances its knowledge progressively, without forgetting previously acquired information.

5. Click on the Save button in the lower right corner of the page.

Note

Implemented classifier should have a similar interface to those provided by the Scikit-learn library. Please do not change the class declaration. Selecting at least two classifiers is required.

Set up a custom metric

1. Click on the Metrics navigation button in the left menu.

2. Click on the Add custom metric button at the top of the page.

3. In the Name section, enter the name of your metric. It should be unique.

4. Choose the type of the metric in section Type.

   • Batch learning– metric prepared for non-incremental datasets.

   • Incremental learning - metric prepared for incremental datasets.

5. Click on the Save button in the lower right corner of the page.

Note

Implemented evaluation metrics should have a similar interface to those provided by the Scikit-learn library. Please do not change the function declaration. Selecting at least one metric is required.

Create a new experiment

1. Click on the Create new experiment button in the lower left corner of the page.

2. In the Name section, enter the name of your experiment. It should be unique.

3. Choose the type of the experiment in the Type section.

   • Batch learning – prepare an experiment for machine learning models learned in a batch manner.

   • Incremental learning - model learns and enhances its knowledge progressively.

4. Select datasets for your experiment.

   • If you want to add a custom dataset, click on the Add custom dataset button and select the appropriate dataset in the modal window.

   • If you want to add a KEEL dataset, click on the Add KEEL dataset button and select the appropriate dataset in the modal window.

   • If you creating an incremental learning experiment, you can also add dataset from River clicking on Add River dataset.

   You can add all datasets at once by clicking on Submit all using default properties or add them one at a time. It is also possible to remove the selected dataset from the Selected datasets section clicking on the Delete button.

5. Select classifiers for your experiment.

   • If you want to add custom classifier, click on the Add custom classifiers button and select the appropriate classifier in the modal window.

   • If you creating a batch learning experiment, you can choose classifiers from the Scikit-learn library for your comparisons. Click on the Add Sklearn classifier button and select the appropriate classifier in the modal window.

   • If you creating an incremental learning experiment, you can choose classifiers from the River library for your comparisons. Click on the Add River classifier button and select the appropriate classifier in the modal window.

   You can add all classifiers at once by clicking on Submit all using default properties or add them one at a time adjust its parameters individually. It is also possible to remove the selected classifier from the Selected classifiers section clicking on the Delete button.

6. Select metrics for your experiment.

   • If you want to add custom metric, click on the Add custom metric button and select the appropriate metric in the modal window.

   • If you creating a batch learning experiment, you can choose metrics from the Scikit-learn library for your comparisons. Click on the Add Sklearn metric button and select the appropriate metric in the modal window.

   • If you creating an incremental learning experiment, you can choose metrics from the River library for your comparisons. Click on the Add River metric button and select the appropriate metric in the modal window.

   You can add all metrics at once by clicking on Submit all using default properties or add them one at a time. It is also possible to remove the selected metric from the Selected metrics section using the Delete button.

7. Click on the Create button and wait for the experiment to complete.

8. Once your experiment is complete, you can view the results by clicking on the Experiment navigation button in the left menu and clicking on the Open button in the row of the selected experiment. To delete the results of a performed experiment, click on the Delete button on the selected experiment.