If data is too large, this class can iteratively absorb data

iterative_hist.py

import numpy as np
import matplotlib.pyplot as plt

class Iterative_hist(object):
    '''
    Collect distribution information in iterative manner. 
    Useful when whole data not fit memory.
    Usage:
        if min and max data value is not known, call set_min_max() method, iterative, on hall dataset
        call add_data() method, for data addition
    '''
    def __init__(self, bins=1000, min_value=None, max_value=None):
        self.min_value = min_value
        self.max_value = max_value
        if min_value and max_value:
            self.hist_bins = np.linspace(self.min, self.max+(self.max-self.min)/self.bins, self.bins)
        self.bins = bins
        self.hist = np.zeros(bins)
#         self.state = False #not updates
        
    def update_min_max(self, data):
        '''
        update data minimum and maximum values
        '''
        min_value = np.min(data)
        max_value = np.max(data)
        if not self.min_value or min_value < self.min_value:
            self.min_value = min_value
        if not self.max_value or max_value > self.max_value:
            self.max_value = max_value
#         self.state = True #updated
        
    def update_hist(self, data):
        '''
        update histogramm
        '''
        assert self.min_value is not None and self.max_value is not None, ' Minimum and maximum values should be defined, given {}, {}. Call update_min_max()'.format(self.min_value, self.max_value)
#         if self.state:
        self.hist_bins = np.linspace(self.min_value, self.max_value+(self.max_value-self.min_value)/self.bins, self.bins)
#             self.state = False
        self.hist += np.bincount(np.digitize(data, self.hist_bins), minlength=self.bins)
        
    def plot(self, log_scale=False):
        plt.bar(self.hist_bins, self.hist, width=self.hist_bins[1]-self.hist_bins[0])
        if log_scale:
            plt.yscale('log')
        plt.show()