Test and summary stats, BY model



In [1]:

    
import matplotlib.pyplot as plt
import pandas as pd



In [2]:

    
%run src/by_monte_carlo_test.py



In [3]:

    
b = BY()



In [4]:

    
by_compute_stat = by_function_factory(b, parallelization_flag=True)



In [5]:

    
by_compute_stat()









    Out[5]:





0.9980300945320473



In [6]:

    
by_compute_stat()









    Out[6]:





0.9980347108148537



In [15]:

    
num_reps = 1000

Varying n



In [16]:

    
vals = np.empty(num_reps)
for i in range(num_reps):
    vals[i] = by_compute_stat(n=500, m=1000)

v = pd.Series(vals)

v.describe()









    Out[16]:





count    1000.000000
mean        0.998086
std         0.000059
min         0.997823
25%         0.998059
50%         0.998097
75%         0.998128
max         0.998209
dtype: float64



In [17]:

    
vals = np.empty(num_reps)
for i in range(num_reps):
    vals[i] = by_compute_stat(n=1000, m=1000)

v = pd.Series(vals)

v.describe()









    Out[17]:





count    1000.000000
mean        0.998128
std         0.000053
min         0.997823
25%         0.998099
50%         0.998136
75%         0.998166
max         0.998240
dtype: float64



In [18]:

    
vals = np.empty(num_reps)
for i in range(num_reps):
    vals[i] = by_compute_stat(n=1500, m=1000)

v = pd.Series(vals)

v.describe()









    Out[18]:





count    1000.000000
mean        0.998159
std         0.000047
min         0.997919
25%         0.998134
50%         0.998167
75%         0.998192
max         0.998246
dtype: float64



In [19]:

    
vals = np.empty(num_reps)
for i in range(num_reps):
    vals[i] = by_compute_stat(n=2000, m=1000)

v = pd.Series(vals)

v.describe()









    Out[19]:





count    1000.000000
mean        0.998149
std         0.000021
min         0.997985
25%         0.998149
50%         0.998157
75%         0.998159
max         0.998160
dtype: float64



In [20]:

    
vals = np.empty(num_reps)
for i in range(num_reps):
    vals[i] = by_compute_stat(n=2500, m=1000)

v = pd.Series(vals)

v.describe()









    Out[20]:





count    1000.000000
mean        0.998208
std         0.000042
min         0.997935
25%         0.998188
50%         0.998214
75%         0.998238
max         0.998288
dtype: float64



In [ ]:

Varying m



In [21]:

    
vals = np.empty(num_reps)
for i in range(num_reps):
    vals[i] = by_compute_stat(n=1000, m=500)

v = pd.Series(vals)

v.describe()









    Out[21]:





count    1000.000000
mean        0.998141
std         0.000058
min         0.997870
25%         0.998112
50%         0.998150
75%         0.998181
max         0.998258
dtype: float64



In [22]:

    
vals = np.empty(num_reps)
for i in range(num_reps):
    vals[i] = by_compute_stat(n=1000, m=1000)

v = pd.Series(vals)

v.describe()









    Out[22]:





count    1000.000000
mean        0.998124
std         0.000054
min         0.997788
25%         0.998096
50%         0.998132
75%         0.998162
max         0.998241
dtype: float64



In [23]:

    
vals = np.empty(num_reps)
for i in range(num_reps):
    vals[i] = by_compute_stat(n=1000, m=2000)

v = pd.Series(vals)

v.describe()









    Out[23]:





count    1000.000000
mean        0.998045
std         0.000023
min         0.997799
25%         0.998044
50%         0.998053
75%         0.998057
max         0.998060
dtype: float64

Large m, n



In [24]:

    
vals = np.empty(num_reps)
for i in range(num_reps):
    vals[i] = by_compute_stat(n=2000, m=2000)

v = pd.Series(vals)

v.describe()









    Out[24]:





count    1000.000000
mean        0.998150
std         0.000026
min         0.997967
25%         0.998148
50%         0.998160
75%         0.998165
max         0.998167
dtype: float64



In [ ]: