Insufficient or Low-Quality Data

ML requires adequate, representative data; with too little or unreliable data, statistical methods or heuristics are safer.