Abstract: A method of training a Machine Learning Algorithm, the method comprising: generating a training set for training the MLA, the training set comprising a plurality of feature vectors and respectively associated estimation errors, the generating comprises: generating the plurality of feature vectors based on history data associated with an industrial process; identifying a respective value of a target process feature within each feature vector; computing a regression function; determining an estimated outcome value for each respective value of the target process feature based on the regression function; and computing the estimation error for each feature vector based on the respectively associated outcome value and the respectively associated estimated outcome value.