sonar_tutorial_part2

How to make predictions

Now let's backtrack a little bit. Somehow we have obtained an interesting individual, the accuracy and/or other performance measurements are promising and we would like to use it to make predictions.

Example examples/sonar02.cc explains how to proceed:

const auto oracle(s.oracle(result.best_individual));
const auto example(random::element(prob.data.selected()));
const auto prediction(oracle->tag(example.input));

std::cout << "Correct class: " << src::label(example)
          << "   Prediction: " << prediction.label
          << "   Sureness: " << prediction.sureness << '\n';

The ultra::src::search class provides the oracle(individual) member function that returns a functor exploitable for symbolic regression and classification tasks.

The key definition is:

const auto oracle(s.oracle(result.best_individual));

The oracle object is a std::unique_ptr<some strange object> smart pointer but this isn't very important. Rather it accepts an example (src::example) and gives the predicted class ((*oracle)(example)).

Even better it has the tag member functions:

const auto prediction(oracle->tag(example.input));

prediction is a struct containing the predicted class (prediction.label) and the sureness of the prediction (prediction.sureness which varies in the [0, 1] interval):

std::cout << "Correct class: " << src::label(example)
          << "   Prediction: " << prediction.label
          << "   Sureness: " << prediction.sureness << '\n';

Model serialization

The model can be made persistent:

std::stringstream ss;
src::serialize::save(ss, oracle);  // save...

// ... and reload it when needed.
const auto oracle2(src::serialize::oracle::load(ss, prob.sset));
const auto prediction2(oracle2->tag(example.input));
std::cout << "   Prediction: " << prediction2.label
          << "   Sureness: " << prediction2.sureness << '\n';

assert(prediction2.label == prediction.label);

PROCEED TO PART 3 →