

In [7]:

    
import numpy as np
import scipy as sp
import scipy.stats as stats
import pandas as pd
import matplotlib.pyplot as plt
import diptest.diptest

%matplotlib inline



In [8]:

    
from numpy.random import choice

Test diptest on simulated distributions



In [3]:

    
dat = np.random.normal(size=200)
dat2 = np.random.normal(loc=4, size=200)
datC = np.concatenate([dat, dat2])



In [4]:

    
fig, (ax1, ax2) = plt.subplots(2, 1, figsize=(10, 10))
ax1.hist(dat)
ax2.hist(datC)









    Out[4]:





(array([  1.,   3.,  33.,  79.,  65.,  36.,  58.,  83.,  33.,   9.]),
 array([-4.0619644 , -2.95400125, -1.8460381 , -0.73807494,  0.36988821,
         1.47785136,  2.58581451,  3.69377766,  4.80174081,  5.90970397,
         7.01766712]),
 <a list of 10 Patch objects>)



In [5]:

    
print(diptest.diptest.diptest(dat))
kde = sp.stats.gaussian_kde(dat)
x_grid = np.linspace(min(dat), max(dat), 1000)
plt.plot(np.cumsum(kde.evaluate(x_grid)))









    



(0.023592414673270577, 0.6494307479884124)






    



---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-5-52018a1aa20a> in <module>()
      1 print(diptest.diptest.diptest(dat))
----> 2 kde = sp.stats.gaussian_kde(dat)
      3 x_grid = np.linspace(min(dat), max(dat), 1000)
      4 plt.plot(np.cumsum(kde.evaluate(x_grid)))

AttributeError: 'module' object has no attribute 'stats'



In [6]:

    
print(diptest.diptest.diptest(datC))
kde = sp.stats.gaussian_kde(datC)
x_grid = np.linspace(min(datC), max(datC), 1000)
plt.plot(np.cumsum(kde.evaluate(x_grid)))









    



(0.047527049109633755, 2.6724431375635405e-06)






    



---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-6-590f0c99f762> in <module>()
      1 print(diptest.diptest.diptest(datC))
----> 2 kde = sp.stats.gaussian_kde(datC)
      3 x_grid = np.linspace(min(datC), max(datC), 1000)
      4 plt.plot(np.cumsum(kde.evaluate(x_grid)))

AttributeError: 'module' object has no attribute 'stats'

Try it on real data



In [26]:

    
import tapeAnalyst.analysis as analysis
import tapeAnalyst.gel_processing as gp



In [27]:

    
dfS = pd.read_csv('/Users/fearjm/devel/tapeAnalyst/data/JH_sample.csv', encoding='ISO-8859-1')
dfS.columns = ['wellID', 'conc', 'description', 'alert', 'notes']



In [129]:

    
gel = gp.TapeStationGel('/Users/fearjm/devel/tapeAnalyst/data/JH_gel.png', dfS)



In [311]:

    
lane = gel.lanes[13]

sec = lane.laneMean[140:400]
smooth = sp.signal.convolve(sec, sp.signal.hamming(10))

x_grid = np.arange(0, smooth.shape[0])
norm = smooth / smooth.sum()
sim = np.random.choice(x_grid, size=3000, replace=True, p=norm)

print(analysis.diptest(sec))

fig, (ax1, ax2) = plt.subplots(2, 1)
ax1.plot(sec)
ax2.plot(smooth)









    



(0.03461538461538462, 0.028267095333036063)






    Out[311]:





[<matplotlib.lines.Line2D at 0x136f35d90>]



In [ ]:



In [14]:

    
np.sort(norm)









    



---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
<ipython-input-14-ee61729e0b1b> in <module>()
----> 1 np.sort(norm)

NameError: name 'norm' is not defined



In [ ]:

    
diptest.diptest.CV?



In [ ]:

    
diptest.diptest._dip(sec)



In [ ]:

    
sp.random.sample?



In [ ]:

    
plt.plot(sec)



In [ ]:

    
myCDF = np.cumsum(sec)
plt.plot(myCDF)



In [ ]:

    
invCDF = 1 / myCDF
plt.plot(invCDF)



In [ ]:

    
invCDF



In [ ]:

    
plt.plot(x_grid, sp.stats.norm.cdf(x_grid))



In [15]:

    
plt.plot(myCDF / myCDF.max())









    



---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
<ipython-input-15-94a51b3e4f4c> in <module>()
----> 1 plt.plot(myCDF / myCDF.max())

NameError: name 'myCDF' is not defined



In [16]:

    
plt.plot(sec)









    Out[16]:





[<matplotlib.lines.Line2D at 0x11d643dd0>]



In [17]:

    
x_grid = np.linspace(0, 250, 10000)



In [18]:

    
x_grid = np.arange(0, len(sec))



In [24]:

    
xs = np.random.choice(x_grid, size=1000000, replace=True, p=sec / sec.sum())



In [25]:

    
kde = sp.stats.gaussian_kde(xs)
plt.plot(kde.evaluate(x_grid))









    Out[25]:





[<matplotlib.lines.Line2D at 0x112130e90>]



In [ ]:



In [ ]: