1. learn
  2. time.ipynb

Predicting DWPC Query runtime ahead of time

In [8]:
import json

import matplotlib.pyplot
import pandas
import numpy
import seaborn
import mpld3

%matplotlib inline
In [9]:
path = 'data/all-features/metapaths.json'
with open(path) as fp:
    metapaths = json.load(fp)
In [10]:
auroc_df = pandas.read_table('data/all-features/auroc.tsv')
auroc_df.head(2)
Out[10]:
metapath nonzero seconds_per_query auroc auroc_permuted delta_auroc pval_auroc length
0 CbGaD 0.312 0.0145 0.715 0.580 0.13500 0.000003 2
1 CbGdD 0.149 0.0136 0.512 0.515 -0.00332 0.921000 2
In [11]:
cols = ['sequential_complexity', 'optimal_join_complexity', 'midpoint_join_complexity']

rows = [[item['abbreviation']] + [item[col] for col in cols] for item in metapaths]
complexity_df = pandas.DataFrame(rows, columns=['metapath'] + cols)
complexity_df = auroc_df.merge(complexity_df)
complexity_df['log10_seconds_per_query'] = numpy.log10(complexity_df['seconds_per_query'])
In [12]:
complexity_df.head(2)
Out[12]:
metapath nonzero seconds_per_query auroc auroc_permuted delta_auroc pval_auroc length sequential_complexity optimal_join_complexity midpoint_join_complexity log10_seconds_per_query
0 CbGaD 0.312 0.0145 0.715 0.580 0.13500 0.000003 2 0.620478 0.713766 0.876638 -1.838632
1 CbGdD 0.149 0.0136 0.512 0.515 -0.00332 0.921000 2 1.206737 0.966103 0.966103 -1.866461

sequential_complexity

In [13]:
matplotlib.pyplot.figure(figsize=(10, 7))
ax = seaborn.regplot('sequential_complexity', 'log10_seconds_per_query', data=complexity_df,
    lowess=True, scatter_kws={'alpha': 0.5}, line_kws={'color': 'black'}, ci=False)
points = ax.collections[0]
labels = complexity_df.metapath.tolist()
tooltip = mpld3.plugins.PointLabelTooltip(points, labels)
mpld3.plugins.connect(ax.figure, tooltip)
mpld3.display()
Out[13]:

optimal_join_complexity

In [14]:
matplotlib.pyplot.figure(figsize=(10, 7))
ax = seaborn.regplot('optimal_join_complexity', 'log10_seconds_per_query', data=complexity_df,
    lowess=True, scatter_kws={'alpha': 0.5}, line_kws={'color': 'black'}, ci=False)
points = ax.collections[0]
labels = complexity_df.metapath.tolist()
tooltip = mpld3.plugins.PointLabelTooltip(points, labels)
mpld3.plugins.connect(ax.figure, tooltip)
mpld3.display()
Out[14]:

midpoint_join_complexity

In [15]:
matplotlib.pyplot.figure(figsize=(10, 7))
ax = seaborn.regplot('midpoint_join_complexity', 'log10_seconds_per_query', data=complexity_df,
    lowess=True, scatter_kws={'alpha': 0.5}, line_kws={'color': 'black'}, ci=False)
points = ax.collections[0]
labels = complexity_df.metapath.tolist()
tooltip = mpld3.plugins.PointLabelTooltip(points, labels)
mpld3.plugins.connect(ax.figure, tooltip)
mpld3.display()
Out[15]: