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?
