Variance is defined as a model’s sensitivity to fluctuations in the data. During training, a model is allowed to ‘see’ data a certain number of times to find patterns in it. If it does not work on the data for long enough, it will not find patterns and bias occurs. On the other hand, if a model is allowed to view the data too many times, it will learn very well for only that data. A model with high variance, however, will have the flexibility to match any data set that’s provided to it, potentially resulting in dramatically different models each time. Variance comes from models that are highly complex, employing a significant number of features (Grant 2019).
Grant, Peter. 2019. “Introducing Model Bias and Variance.” Towards Data Science. 2019. https://towardsdatascience.com/introducing-model-bias-and-variance-187c….