Baselining with AutoML
A baseline model is a simple model used as a starting point for ML. Data scientists often use a baseline model to compare the performance of more complex models. Baseline models are typically simple or common algorithms, such as the majority class classifier or a random forest.
Baseline models are valuable for several reasons, some of which are listed here:
- They can help you understand the difficulty of finding a signal given your current dataset. If even the best baseline model performs poorly, it may indicate that more complex models will also struggle to find useful patterns (that is, garbage data in, garbage models out).
- Baseline models can help you to identify features that are most important for the ML task. If a baseline model performs well, it may be because it can learn from the most salient features.
- Baseline models can help you avoid overfitting. Overfitting is a frequent problem with more complex models. It occurs when a model learns...
