14(Burkov, Machine Learning Engineering)
15Note this is different from a biased model. Bias also introduces error through the inability to express the functional relationship between the input and output data, however this is due to strong prior assumptions about the data’s functional form. This is an important consideration when using a parametrized model like linear regression. More data will not directly reduce error from bias, but more data does enable the use of more flexible parametrized algorithms for which model bias can be reduced completely.
16(Burkov, The Hundred-Page Machine Learning Book 97 - 98)