Пример #1
0
    def __init__(self, f_weight, pars_w, f_dens, pars_d):
        """The interpolator container. It is the minimum container for
        interpolation tasks.

        Parameters
        ----------
        f_weight: function
            the function which defines the weights from distances.
        pars_w: dict
            the parameters of the weights definition from distances.
        f_dens: function
            the function definition of the density assignation kernel.
        pars_d: dict
            the parameters for assigning the kernel.

        """
        characterizer = create_weighted_function(f_weight, pars_w,
                                                 f_dens, pars_d)
        ## Specific class settings
        self.compute = characterizer
        ## Initial function set
        self._out_formatter = null_out_formatter
        self._f_default_names = array_featurenames
        self._defult_add2result = sum_addresult_function
        self._assert_correctness()
def test():
    n = 100
    locs = np.random.random((n, 2))*100
    locs1 = random_transformed_space_points(n/10, 2, None)*100
    feats0 = np.random.random((n/10, 4))
    feats1 = np.random.random(n/10)
    kneighs4 = np.ones(len(locs)).astype(int)*4
    ret = KRetriever(locs1, 4, ifdistance=True)
    ret1 = KRetriever(locs1, ifdistance=True)

    f_dens0, f_dens1, f_dens2 = 'weighted_count', 'weighted_avg', 'null'
    pars_d = {}
    f_weights = ['null', 'linear', 'Trapezoid', 'inverse_prop', 'inverse_prop',
                 'exponential', 'exponential', 'gaussian', 'gaussian',
                 'surgaussian', 'surgaussian', 'surgaussian', 'sigmoid',
                 'sigmoid']

    pars_w1 = {'max_r': 0.2, 'max_w': 1, 'min_w': 0}
    pars_w2 = {'max_r': 0.2, 'r2': 0.4, 'max_w': 1, 'min_w': 0}

    pars_w3 = {'max_r': 0.2, 'max_w': 1, 'min_w': 1e-8, 'rescale': True}
    pars_w3b = {'max_r': 0.2, 'max_w': 1, 'min_w': 1e-8, 'rescale': False}
    pars_w5 = {'max_r': 0.2, 'max_w': 1, 'min_w': 0, 'S': None,
               'rescale': True}
    pars_w5b = {'max_r': 0.2, 'max_w': 1, 'min_w': 1e-3, 'S': None,
                'rescale': False}
    pars_w6 = {'max_r': 0.2, 'max_w': 1, 'min_w': 1e-3, 'S': 0.5,
               'rescale': True}
    pars_w7 = {'max_r': 0.2, 'max_w': 1, 'min_w': 1e-3, 'r_char': 0, 'B': None,
               'rescale': True}
    pars_w7b = {'max_r': 0.2, 'max_w': 1, 'min_w': 1e-3, 'r_char': 0, 'B': 2,
                'rescale': False}
    pars_w = [{}, pars_w1, pars_w2, pars_w3, pars_w3b, pars_w3, pars_w3b,
              pars_w5, pars_w5b, pars_w5, pars_w5b, pars_w6, pars_w7, pars_w7b]

    values, dists = np.random.random((10, 4)), np.random.random(10)

    pars_now = {'max_r': 0.2, 'max_w': 1, 'min_w': 1e-3, 'r_char': 0}
    set_scales_kernel('surgaussian', **pars_now)
    set_scales_kernel('gaussian', **pars_now)
    set_scales_kernel('sigmoid', **pars_now)

    for i in range(len(f_weights)):
        f = create_weighted_function(f_weights[i], pars_w[i], f_dens0, pars_d)
        f(values, dists)
        f = create_weighted_function(f_weights[i], pars_w[i], f_dens1, pars_d)
        f(values, dists)
        f = create_weighted_function(f_weights[i], pars_w[i],
                                     characterizer_summer, pars_d)
        f(values, dists)

    M = general_density_assignation(locs, ret, kneighs4, feats0,
                                    f_weights[1], pars_w[1], f_dens1, pars_d)
    M = general_density_assignation(locs, ret, kneighs4, feats1,
                                    f_weights[1], pars_w[1], f_dens1, pars_d)
    M = general_density_assignation(locs, ret1, kneighs4, feats1,
                                    f_weights[1], pars_w[1], f_dens1, pars_d)
    M = general_density_assignation(locs, ret1, kneighs4, feats1,
                                    f_weights[1], pars_w[1], f_dens0, pars_d)

    interpolator = Interpolator(f_weights[1], pars_w[1], f_dens1, pars_d)

    ###########################################################################
    ###########################################################################
    #### Density assignation testing
    ################################
    logfile = Logger('logfile.log')
    pars_dens_asign = {'f_weights': f_weights[1], 'params_w': pars_w[1],
                       'f_dens': f_dens1, 'params_d': pars_d}
    locs_data = locs
    pop_data = feats1
    retriever = KRetriever
    dens_asign = DensityAssign_Process(logfile, retriever)
    dens_asign.compute_density(locs1, locs_data, pop_data, kneighs4,
                               pars_dens_asign)
    os.remove('logfile.log')

    ###########################################################################
    ###########################################################################
    #### Interpolation utils testing
    ################################
    density0 = np.random.random(50)*2
    density1 = np.random.random(50)*2
    clustering_by_comparison(density0, density1)