huRandom Tools

data_compensation.py

#!/usr/bin/python
# This compensates for troll data in which a user decides to enter the same value a few, possibly one hundred,
# times to hurandom.elijahciane.me
import pygal
from pygal.style import CleanStyle

data = [-100000, -100, -100, 0, 0.4919173444481295, 1, 1, 1, 2, 3, 4, 4, 5, 5, 6, 7, 8, 8, 9, 9, 10, 12, 13, 13, 13, 14, 18, 20, 20, 22, 22, 22, 22, 23, 25, 29, 32, 32, 37, 37, 37, 37, 40, 42, 42, 42, 42, 42, 42, 43, 55, 64, 65, 67, 67, 71, 76, 79, 82, 97.00027461847252, 123, 137, 164, 337, 337.2346435765485, 365, 437, 456, 558, 632, 658, 666, 789, 1764, 3466, 3788, 3857, 4550, 5672, 5761, 9586, 10021, 12334, 12789, 12789, 12921, 35614, 44307, 86372, 98776, 114413, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135484, 135485, 135486, 135489, 135490, 135490, 135490, 135490, 135490, 135490, 141928, 233500, 259152, 325756, 345235, 422749, 684613, 696969, 764509] 

print('\nData: ' + str(data))

new_data = list()

for i in range(1, len(data)-1):
    if data[i-1] != data[i+1]:
        new_data.append(data[i])

print('\nNew Data: ' + str(new_data))

def graph_data(data):
    """
    Generates histogram of data for user
    """
    graph_data, axis , n= compile_data(data)
    chart = pygal.Bar(fill=True,
                      interpolate='cubic',
                      style=CleanStyle,
                      no_data_text=str(n) + ' more entries needed for graphing.',
                      title_font_size=30,
                      show_legend=False,
                      x_title='Range',
                      y_title='Frequency',
                      label_font_size=12.5,
                      x_label_rotation=30,
                      title="'Random' Numbers",
                      x_labels=axis,
                      no_data_font_size=30)
    chart.add('Data', graph_data)
    chart.render_to_file('bar_chart.svg')

def compile_data(data):
    """
    Compiles data for graphing
    Splits the data evenly into #bin_num of 'bins' which store the frequency of
    number occurences within each range.
    Breaks if you enter a number that is magnitudes larger than the rest of
    the data.
    """
    bin_num = 20
    bin_range = int((max(data) - min(data)) / bin_num)+1
    output = [0 for i in range(bin_num)]
    axis = ['-' for i in range(bin_num)]
    n = bin_num - len(data)

    # Purely debigging information
    #print('Number of Bins: ' + str(bin_num) +\
    #      ' | range for each bin: ' + str(bin_range) +\
    #      ' | min value: ' + str(min(data)) +\
    #      ' | max value: ' + str(max(data)) +\
    #      ' | number of values: ' + str(len(data)) +\
    #      ' | list of data: ' + str(data))

    if len(data) >= bin_num:
        for i in range(bin_num):
            axis[i] = str(bin_range*i+min(data))
            for j in range(len(data)):
                if bin_range*i+min(data) <= data[j] and data[j] < bin_range*(i+1)+min(data):
                    output[i] += 1
            axis[i] += " to " + str(bin_range*(i+1)+min(data))

    return(output, axis, n)

graph_data(new_data)