Bins Analysis with Numpy

bin_study.py

import pickle as pkl
import numpy as np
import seaborn as sns

# load data
scores = pkl.load(open('y_scores_20.pkl','rb'))

# create bins
bins = np.linspace(0, 1, 157) # 157 is determined arbitrarily 
digitized = np.digitize(scores, bins)
bin_means = [scores[digitized == i].mean() for i in range(1, len(bins))]

# get the histogram, https://docs.scipy.org/doc/numpy/reference/generated/numpy.histogram.html

hist =  np.histogram(scores,bins='sturges')
np.savetxt('bins.csv',scores[scores > hist[1][1]],delimiter=';') # export without the first bin

# if visualized in tableau, bins are computed as follows: Number of Bins = 3 + log2(n) * log(n)
# ref: http://onlinehelp.tableau.com/current/pro/desktop/en-us/calculations_bins.html

# plot with seaborn
dist = sns.distplot(scores[scores > hist[1][1]]) # visualize without the first bin
dist.figure.savefig('output.png')