I am training my data using following code:
start_time := clock_timestamp(); PERFORM madlib.create_nb_prepared_data_tables( 'nb_training', 'class', 'attributes', 'ARRAY[1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57]', 57, 'categ_feature_probs', 'numeric_attr_params', 'class_priors' ); training_time := 1000* (extract(epoch FROM clock_timestamp()) - extract(epoch FROM start_time)); And my prediction code goes as follows: start_time := clock_timestamp(); PERFORM madlib.create_nb_probs_view( 'categ_feature_probs', 'class_priors', 'nb_testing', 'id', 'attributes', 57, 'numeric_attr_params', 'probs_view' ); select * from probs_view prediction_time := 1000 * (extract(epoch FROM clock_timestamp()) - extract(epoch FROM start_time)); The training data is containing 450000 records were as testing dataset contains 50000 records. Still, my average training_time is around 17173 ms where as prediction_time is 26481 ms. As per my understanding of naive bayes, the prediction_time should be less than training_time. What am I doing wrong here?