Data Analysis for AM 205 Project

Wesley Chen, Brandon Sim, Andy Shi



In [1]:

    
import numpy as np
import matplotlib.pyplot as plt
% matplotlib inline
import pandas as pd
import itertools
import seaborn

plt.rcParams['figure.figsize'] = (10, 8)



In [2]:

    
data1 = pd.read_csv("data1.csv")
data1['Example'] = 1
data1.head()









    Out[2]:






  
    
      
      n_procs
      numSteps
      qualityfactor
      corrections
      serial_max
      parallel_max
      serial_mean
      parallel_mean
      serial_time
      parallel_time
      Example
    
  
  
    
      0
        2
       100
       100
       2
       0.000062
       0.00615
       0.000026
       0.002623
       0.144839
       7.290737
       1
    
    
      1
        4
       100
        10
       2
       0.000619
       0.00615
       0.000264
       0.002623
       0.016247
       4.734909
       1
    
    
      2
        4
       100
        50
       2
       0.000124
       0.00615
       0.000053
       0.002623
       0.068887
       5.979487
       1
    
    
      3
        8
       100
       100
       2
       0.000062
       0.00615
       0.000026
       0.002623
       0.226073
       3.781236
       1
    
    
      4
       16
       100
        10
       2
       0.000619
       0.00615
       0.000264
       0.002623
       0.082640
       5.872348
       1



In [3]:

    
data2 = pd.read_csv("data2.csv")
data2['Example'] = 2
data2.head()









    Out[3]:






  
    
      
      n_procs
      numSteps
      qualityfactor
      corrections
      serial_max
      parallel_max
      serial_mean
      parallel_mean
      serial_time
      parallel_time
      Example
    
  
  
    
      0
        2
       100
       100
       2
       0.000353
       0.037062
       0.000163
       0.017199
       0.273088
       15.602692
       2
    
    
      1
        4
       100
        10
       2
       0.003544
       0.037062
       0.001641
       0.017199
       0.034686
       10.134147
       2
    
    
      2
        4
       100
        50
       2
       0.000706
       0.037062
       0.000327
       0.017199
       0.131030
        9.705749
       2
    
    
      3
        8
       100
       100
       2
       0.000353
       0.037062
       0.000163
       0.017199
       0.323222
        6.800989
       2
    
    
      4
       16
       100
        10
       2
       0.003544
       0.037062
       0.001641
       0.017199
       0.095807
       13.253634
       2



In [4]:

    
data = pd.concat([data1, data2], axis=0, ignore_index=True)
data









    Out[4]:






  
    
      
      n_procs
      numSteps
      qualityfactor
      corrections
      serial_max
      parallel_max
      serial_mean
      parallel_mean
      serial_time
      parallel_time
      Example
    
  
  
    
      0 
        2
       100
       100
         2
       0.000062
       0.006150
       0.000026
       0.002623
       0.144839
         7.290737
       1
    
    
      1 
        4
       100
        10
         2
       0.000619
       0.006150
       0.000264
       0.002623
       0.016247
         4.734909
       1
    
    
      2 
        4
       100
        50
         2
       0.000124
       0.006150
       0.000053
       0.002623
       0.068887
         5.979487
       1
    
    
      3 
        8
       100
       100
         2
       0.000062
       0.006150
       0.000026
       0.002623
       0.226073
         3.781236
       1
    
    
      4 
       16
       100
        10
         2
       0.000619
       0.006150
       0.000264
       0.002623
       0.082640
         5.872348
       1
    
    
      5 
       16
       100
        50
         2
       0.000124
       0.006150
       0.000053
       0.002623
       0.144870
         2.524051
       1
    
    
      6 
       16
       100
       100
         2
       0.000062
       0.006150
       0.000026
       0.002623
       0.144561
         4.177090
       1
    
    
      7 
       16
       100
       100
         1
       0.000062
       0.006243
       0.000026
       0.002663
       0.884363
         0.024045
       1
    
    
      8 
       16
       100
       100
         2
       0.000062
       0.006150
       0.000026
       0.002623
       0.476075
         1.524805
       1
    
    
      9 
       16
       100
       100
         5
       0.000062
       0.006243
       0.000026
       0.002663
       0.138123
         4.277979
       1
    
    
      10
       16
       100
       100
        10
       0.000062
       0.006150
       0.000026
       0.002623
       0.276352
         9.230153
       1
    
    
      11
       16
       100
       100
        50
       0.000062
       0.006150
       0.000026
       0.002623
       0.307629
        48.117501
       1
    
    
      12
       16
       100
       100
       100
       0.000062
       0.006150
       0.000026
       0.002623
       0.518091
       105.657221
       1
    
    
      13
       32
       100
        10
         2
       0.000619
       0.006150
       0.000264
       0.002623
       0.024898
         2.224934
       1
    
    
      14
       32
       100
        50
         2
       0.000124
       0.006150
       0.000053
       0.002623
       0.071008
         2.119910
       1
    
    
      15
       32
       100
       100
         1
       0.000062
       0.006243
       0.000026
       0.002663
       0.443254
         0.047888
       1
    
    
      16
       32
       100
       100
         2
       0.000062
       0.006150
       0.000026
       0.002623
       0.119375
         1.640201
       1
    
    
      17
       32
       100
       100
         5
       0.000062
       0.006243
       0.000026
       0.002663
       0.309229
         7.003868
       1
    
    
      18
       32
       100
       100
        10
       0.000062
       0.006150
       0.000026
       0.002623
       0.263915
        14.185466
       1
    
    
      19
       32
       100
       100
        50
       0.000062
       0.006150
       0.000026
       0.002623
       0.279576
        72.412345
       1
    
    
      20
       32
       100
       100
       100
       0.000062
       0.006150
       0.000026
       0.002623
       0.298442
       161.320850
       1
    
    
      21
        2
       100
       100
         2
       0.000353
       0.037062
       0.000163
       0.017199
       0.273088
        15.602692
       2
    
    
      22
        4
       100
        10
         2
       0.003544
       0.037062
       0.001641
       0.017199
       0.034686
        10.134147
       2
    
    
      23
        4
       100
        50
         2
       0.000706
       0.037062
       0.000327
       0.017199
       0.131030
         9.705749
       2
    
    
      24
        8
       100
       100
         2
       0.000353
       0.037062
       0.000163
       0.017199
       0.323222
         6.800989
       2
    
    
      25
       16
       100
        10
         2
       0.003544
       0.037062
       0.001641
       0.017199
       0.095807
        13.253634
       2
    
    
      26
       16
       100
        50
         2
       0.000706
       0.037062
       0.000327
       0.017199
       0.186552
         5.194865
       2
    
    
      27
       16
       100
       100
         2
       0.000353
       0.037062
       0.000163
       0.017199
       0.379339
         9.783687
       2
    
    
      28
       16
       100
       100
         1
       0.000353
       0.037062
       0.000163
       0.017205
       0.304592
         0.033993
       2
    
    
      29
       16
       100
       100
         2
       0.000353
       0.037062
       0.000163
       0.017199
       0.300513
         2.786015
       2
    
    
      30
       16
       100
       100
         5
       0.000353
       0.037062
       0.000163
       0.017205
       0.296380
         8.982083
       2
    
    
      31
       16
       100
       100
        10
       0.000353
       0.037062
       0.000163
       0.017199
       0.315678
        19.845286
       2
    
    
      32
       16
       100
       100
        50
       0.000353
       0.037062
       0.000163
       0.017199
       0.283889
       105.358575
       2
    
    
      33
       16
       100
       100
       100
       0.000353
       0.037062
       0.000163
       0.017199
       0.308455
       230.499720
       2
    
    
      34
       32
       100
        10
         2
       0.003544
       0.037062
       0.001641
       0.017199
       0.062029
         3.863887
       2
    
    
      35
       32
       100
        50
         2
       0.000706
       0.037062
       0.000327
       0.017199
       0.139086
         4.521666
       2
    
    
      36
       32
       100
       100
         1
       0.000353
       0.037062
       0.000163
       0.017205
       0.511949
         0.041012
       2
    
    
      37
       32
       100
       100
         2
       0.000353
       0.037062
       0.000163
       0.017199
       0.273211
         3.680919
       2
    
    
      38
       32
       100
       100
         5
       0.000353
       0.037062
       0.000163
       0.017205
       0.289979
        16.930516
       2
    
    
      39
       32
       100
       100
        10
       0.000353
       0.037062
       0.000163
       0.017199
       0.286186
        30.488317
       2
    
    
      40
       32
       100
       100
        50
       0.000353
       0.037062
       0.000163
       0.017199
       0.269222
       117.171796
       2
    
    
      41
       32
       100
       100
       100
       0.000353
       0.037062
       0.000163
       0.017199
       0.311367
       281.408144
       2



In [5]:

    
def plot_stuff(grouped, x, field1, field2, xlab="", ylab="", title="", fname="plot.pdf"):
    grouped1 = grouped[grouped['Example'] == 1]
    grouped2 = grouped[grouped['Example'] == 2]
    plt.plot(grouped1[x], grouped1[field1], '-o', label='Example 1 Serial')
    plt.plot(grouped1[x], grouped1[field2], '-o', label='Example 1 Parallel')
    plt.plot(grouped2[x], grouped2[field1], '-o', label='Example 2 Serial')
    plt.plot(grouped2[x], grouped2[field2], '-o', label='Example 2 Parallel')
    plt.legend(loc='best', bbox_to_anchor = (1.25,1))
    plt.title(title)
    plt.ylabel(ylab)
    plt.xlabel(xlab)
    plt.savefig(fname, bbox_inches="tight")
    plt.show()



In [6]:

    
# plot of runtime vs CPUs
criterion = data['corrections'].isin([2]) & data['qualityfactor'].isin([100])
plt_data = data[criterion]
grouped = plt_data.groupby(['n_procs', 'Example'], as_index=False).aggregate(np.mean)
plot_stuff(grouped, 'n_procs', 'serial_time', 'parallel_time', 'Number of CPUs', 
           'Wall Clock Time (sec)', 'Runtime vs. CPUs', 'runtime_vs_cpus.pdf')



In [7]:

    
# plot of runtime vs number of corrections
plt_df = data[data['n_procs'].isin([16]) & data['qualityfactor'].isin([100])]
grouped = plt_df.groupby(['corrections', 'Example'], as_index=False).aggregate(np.mean)
plot_stuff(grouped, 'corrections', 'serial_time', 'parallel_time', 'Number of Corrections',
           'Wall Clock Time (sec)', 'Runtime vs. Number of Correction Steps', 'runtime_vs_corrections.pdf')



In [8]:

    
# plot of accuracy vs number of corrections
plt_df = data[data['n_procs'].isin([32]) & data['qualityfactor'].isin([100])]
grouped = plt_df.groupby(['corrections', 'Example'], as_index=False).aggregate(np.mean)
plot_stuff(grouped, 'corrections', 'serial_mean', 'parallel_mean', 'Number of Corrections',
           'Absolute Error', 'Error vs. Number of Correction Steps', 'error_vs_corrections.pdf')



In [9]:

    
# plot of accuracy vs qualityfactor
plt_df = data[data['n_procs'].isin([32]) & data['corrections'].isin([2])]
grouped = plt_df.groupby(['qualityfactor', 'Example'], as_index=False).aggregate(np.mean)
plot_stuff(grouped, 'qualityfactor', 'serial_mean', 'parallel_mean', 'Quality Factor',
           'Absolute Error', 'Error vs. Quality Factor', 'error_vs_qualityfactor.pdf')

	n_procs	numSteps	qualityfactor	corrections	serial_max	parallel_max	serial_mean	parallel_mean	serial_time	parallel_time	Example
0	2	100	100	2	0.000062	0.00615	0.000026	0.002623	0.144839	7.290737	1
1	4	100	10	2	0.000619	0.00615	0.000264	0.002623	0.016247	4.734909	1
2	4	100	50	2	0.000124	0.00615	0.000053	0.002623	0.068887	5.979487	1
3	8	100	100	2	0.000062	0.00615	0.000026	0.002623	0.226073	3.781236	1
4	16	100	10	2	0.000619	0.00615	0.000264	0.002623	0.082640	5.872348	1

	n_procs	numSteps	qualityfactor	corrections	serial_max	parallel_max	serial_mean	parallel_mean	serial_time	parallel_time	Example
0	2	100	100	2	0.000353	0.037062	0.000163	0.017199	0.273088	15.602692	2
1	4	100	10	2	0.003544	0.037062	0.001641	0.017199	0.034686	10.134147	2
2	4	100	50	2	0.000706	0.037062	0.000327	0.017199	0.131030	9.705749	2
3	8	100	100	2	0.000353	0.037062	0.000163	0.017199	0.323222	6.800989	2
4	16	100	10	2	0.003544	0.037062	0.001641	0.017199	0.095807	13.253634	2

	n_procs	numSteps	qualityfactor	corrections	serial_max	parallel_max	serial_mean	parallel_mean	serial_time	parallel_time	Example
0	2	100	100	2	0.000062	0.006150	0.000026	0.002623	0.144839	7.290737	1
1	4	100	10	2	0.000619	0.006150	0.000264	0.002623	0.016247	4.734909	1
2	4	100	50	2	0.000124	0.006150	0.000053	0.002623	0.068887	5.979487	1
3	8	100	100	2	0.000062	0.006150	0.000026	0.002623	0.226073	3.781236	1
4	16	100	10	2	0.000619	0.006150	0.000264	0.002623	0.082640	5.872348	1
5	16	100	50	2	0.000124	0.006150	0.000053	0.002623	0.144870	2.524051	1
6	16	100	100	2	0.000062	0.006150	0.000026	0.002623	0.144561	4.177090	1
7	16	100	100	1	0.000062	0.006243	0.000026	0.002663	0.884363	0.024045	1
8	16	100	100	2	0.000062	0.006150	0.000026	0.002623	0.476075	1.524805	1
9	16	100	100	5	0.000062	0.006243	0.000026	0.002663	0.138123	4.277979	1
10	16	100	100	10	0.000062	0.006150	0.000026	0.002623	0.276352	9.230153	1
11	16	100	100	50	0.000062	0.006150	0.000026	0.002623	0.307629	48.117501	1
12	16	100	100	100	0.000062	0.006150	0.000026	0.002623	0.518091	105.657221	1
13	32	100	10	2	0.000619	0.006150	0.000264	0.002623	0.024898	2.224934	1
14	32	100	50	2	0.000124	0.006150	0.000053	0.002623	0.071008	2.119910	1
15	32	100	100	1	0.000062	0.006243	0.000026	0.002663	0.443254	0.047888	1
16	32	100	100	2	0.000062	0.006150	0.000026	0.002623	0.119375	1.640201	1
17	32	100	100	5	0.000062	0.006243	0.000026	0.002663	0.309229	7.003868	1
18	32	100	100	10	0.000062	0.006150	0.000026	0.002623	0.263915	14.185466	1
19	32	100	100	50	0.000062	0.006150	0.000026	0.002623	0.279576	72.412345	1
20	32	100	100	100	0.000062	0.006150	0.000026	0.002623	0.298442	161.320850	1
21	2	100	100	2	0.000353	0.037062	0.000163	0.017199	0.273088	15.602692	2
22	4	100	10	2	0.003544	0.037062	0.001641	0.017199	0.034686	10.134147	2
23	4	100	50	2	0.000706	0.037062	0.000327	0.017199	0.131030	9.705749	2
24	8	100	100	2	0.000353	0.037062	0.000163	0.017199	0.323222	6.800989	2
25	16	100	10	2	0.003544	0.037062	0.001641	0.017199	0.095807	13.253634	2
26	16	100	50	2	0.000706	0.037062	0.000327	0.017199	0.186552	5.194865	2
27	16	100	100	2	0.000353	0.037062	0.000163	0.017199	0.379339	9.783687	2
28	16	100	100	1	0.000353	0.037062	0.000163	0.017205	0.304592	0.033993	2
29	16	100	100	2	0.000353	0.037062	0.000163	0.017199	0.300513	2.786015	2
30	16	100	100	5	0.000353	0.037062	0.000163	0.017205	0.296380	8.982083	2
31	16	100	100	10	0.000353	0.037062	0.000163	0.017199	0.315678	19.845286	2
32	16	100	100	50	0.000353	0.037062	0.000163	0.017199	0.283889	105.358575	2
33	16	100	100	100	0.000353	0.037062	0.000163	0.017199	0.308455	230.499720	2
34	32	100	10	2	0.003544	0.037062	0.001641	0.017199	0.062029	3.863887	2
35	32	100	50	2	0.000706	0.037062	0.000327	0.017199	0.139086	4.521666	2
36	32	100	100	1	0.000353	0.037062	0.000163	0.017205	0.511949	0.041012	2
37	32	100	100	2	0.000353	0.037062	0.000163	0.017199	0.273211	3.680919	2
38	32	100	100	5	0.000353	0.037062	0.000163	0.017205	0.289979	16.930516	2
39	32	100	100	10	0.000353	0.037062	0.000163	0.017199	0.286186	30.488317	2
40	32	100	100	50	0.000353	0.037062	0.000163	0.017199	0.269222	117.171796	2
41	32	100	100	100	0.000353	0.037062	0.000163	0.017199	0.311367	281.408144	2