makeTrees.py

from abc import ABC, abstractmethod
import time
import multiprocessing
import logging

class MakeTrees(ABC):
    @abstractmethod
    def make_decisions_trees(self,dataset,rf):
        pass


class Multiprocessing(MakeTrees):
    def _target(self, dataset, nproc,rf):
        print('process {} starts'.format(nproc))
        initial_depth = 1
        tree = rf._make_node(dataset, initial_depth)
        print('process {} ends'.format(nproc))
        return tree

    def make_decisions_trees(self,dataset,rf):
        t1 = time.time()
        with multiprocessing.Pool() as pool:
            list_of_trees = pool.starmap(self._target, [(dataset, nprocess,rf) for nprocess in range(rf.num_trees)])
            # use pool.map instead if only one argument for _target
        t2 = time.time()
        print('{} seconds per tree'.format((t2 - t1) / rf.num_trees))
        return list_of_trees


class Lineal(MakeTrees):
    def make_decisions_trees(self,dataset,rf):
        list_of_trees = []
        initial_depth = 1
        t1 = time.time()
        for _ in range(rf.num_trees):
            train_dataset = dataset.select_samples(rf.ratio_samples)
            tree = rf._make_node(train_dataset, initial_depth)
            list_of_trees.append(tree)
        t2 = time.time()
        print('{} seconds per tree'.format((t2 - t1) / rf.num_trees))
        return list_of_trees