Underfitting

What is ML Underfitting?

Model fitting is an important part of machine learning modelling and the accuracy of the particular model. So, what is model underfitting in machine learning?

Underfitting occurs when a machine learning model cannot capture the relationship between the input data and the output target value. Model underfitting is when the model performs poorly on the training data, or both the training and the unseen data. This happens due to a high simplicity of the model – in other words, the machine learning model is too simple.

Reasons for model underfitting:

1. High bias in the model performance
2. High model simplicity
3. Low data availability
4. Low data preparation

The following is an example of underfitting a ML model.

How to tackle model underfitting:

1. Feature engineering
2. Adding more data
3. More efficient data preprocessing
4. Reduction in regularisation
5. Increase in training duration

The following is an example of a balanced machine learning model.

Next: Overfitting