underlying assumptions of ML algorithms

Question

Accepted Answer

ML algorithms assume that the data is representative and unbiased.

ML algorithms assume that the data is accurately labeled and free from errors.
They assume that the data is relevant to the problem being solved.
They assume that the data is representative of the population it was sampled from.
They assume that the data is unbiased and not influenced by any external factors.
ML algorithms also assume that the relationships between the input and output variables are consistent and stable.
For example, a spam filter algorithm assumes that the training data accurately represents the types of emails that users receive.
Another example is a medical diagnosis algorithm that assumes the training data is representative of the population being diagnosed.
ML algorithms also assume that the input data is independent and identically distributed (IID).

Top Deloitte Lead Solution Advisor interview questions & answers