Frequently Asked Questions

1. Why isn't my trial starting?

Trial settings need to be adjusted:

• too many unique classes in the target column
• If the trial is not an NLP trial; deselect the models NLP Sequence Classifier/Regressor from Deep Learning Model section in model selection page.

2. What types of machine learning tasks does evoML support?

evoML supports classification and regression tasks. For classification, it can handle both binary and multi-class problems. We also support timeseries classification and timeseries regression tasks.

3. How does evoML handle imbalanced data?

evoML automatically identifies imbalanced features and allows users to used sampling techniques to determine the best sampling method.

4. Can I customize the models used in evoML?

Yes, evoML allows you to select from a range of available models. You can manually select or deselect models, and choose sets optimized for speed, explainability, or advanced performance.

5. How does evoML ensure model explainability?

evoML provides transparency by making the source code of deployed models available for download. Moreover, we allow users to view details on the encoders used for features.

6. How can I evaluate the performance of my models in evoML?

evoML provides various evaluation metrics and visualizations, including confusion matrices and ROC curve graphs, to help you assess model performance.

7. Is it possible to download the code for my models?

Yes, evoML allows you to download production-ready code for your models.

8. How does evoML optimise models for different hardware configurations?

The evoML platform automatically detects the availability of GPUs. If a GPU is available, evoML will utilize it for optimal performance. Otherwise, it will default to using the CPU.