In [1]:

    

%autosave 0
from datetime import date
import os


    

    















    




Autosave disabled

In [2]:

    

import funcy
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib
import matplotlib.pyplot as plt
%matplotlib inline
plt.style.use(['seaborn-white', 'seaborn-paper'])
matplotlib.rc("font", size=22)


    

In [3]:

    

!ls output/mean/analysis


    

    




counts_mean_pct.tex		planning_times_imputedb.csv
counts_mean.tex			running_times_base_tables_1.csv
counts_std.tex			running_times_base_tables.csv
perf_summary.tex		running_times_imputedb_1.csv
planning_times_base_tables.csv	running_times_imputedb.csv

In [4]:

    

# Load data
def load_data(kind, name, subdir=""):
    df = pd.read_csv(os.path.join("output", subdir, "{}_{}_1.csv".format(kind, name)), index_col=0)
    return df

nqueries = 9
times = {}
for method in ["mean", "hot_deck", "regression_tree"]:
    subdir = os.path.join(method, "analysis")
    times[method] = {}
    for scale in ["imputedb", "base_tables"]:
        times[method][scale] = load_data("running_times", scale, subdir)
        
    times[method]["combined"] = times[method]["base_tables"].append(times[method]["imputedb"])


    

In [5]:

    

# Round alpha to nearest 0.05
def round_alpha(alpha):
    if alpha == "Impute at base tables":
        return "Impute at base tables"
    else:
        return round(alpha * 100 / 5) / (100 / 5)
    
for method in ["mean", "hot_deck", "regression_tree"]:
    for scale in ["imputedb", "base_tables", "combined"]:
        times[method][scale]["alpha"] = times[method][scale]["alpha"].apply(round_alpha)


    

In [6]:

    

times["regression_tree"]["base_tables"].head()


    

    
Out[6]:






  

    

      

      
query
      
alpha
      
mean
      
std
    
  
  

    

      
0
      
0.0
      
Impute at base tables
      
24935.606818
      
5549.724521
    
    

      
1
      
1.0
      
Impute at base tables
      
6553.513636
      
50.731587
    
    

      
2
      
2.0
      
Impute at base tables
      
10650.927273
      
49.561771
    
    

      
3
      
3.0
      
Impute at base tables
      
10652.572727
      
58.008496
    
    

      
4
      
4.0
      
Impute at base tables
      
10646.022727
      
51.340162
    
  

In [7]:

    

times["regression_tree"]["imputedb"].head()


    

    
Out[7]:






  

    

      

      
query
      
alpha
      
mean
      
std
    
  
  

    

      
0
      
0.0
      
0.0
      
708.127273
      
312.209595
    
    

      
1
      
0.0
      
0.5
      
324.595455
      
150.178299
    
    

      
2
      
0.0
      
1.0
      
18.636364
      
5.430828
    
    

      
3
      
1.0
      
0.0
      
556.231818
      
11.419997
    
    

      
4
      
1.0
      
0.5
      
94.227273
      
2.738726
    
  

In [8]:

    

def plot(times, method):
    df = times[method]["combined"].copy()
    kind = "running_times"
    name = "combined"
    
    # plots
    xticks = range(0, nqueries)
    xlabels = ["%i" % (q + 1) for q in xticks]
    
    f = plt.figure()

    df = df.pivot(index="query",columns="alpha",values="mean")
    df = df[["Impute at base tables",0.0,0.5,1.0]]
    df = df.rename(columns={
        0.0: r"$\alpha = 0.0$",
        0.5: r"$\alpha = 0.5$",
        1.0: r"$\alpha=1.0$",
        "Impute at base tables" : "Baseline"
    })
    ax = df.plot(kind="bar")
    ax.set_yscale("log")
    ax.set_ylim(bottom=0.99, top=10e5-0.1)
    ax.legend(title=None, loc="upper left", fontsize=14, ncol=2)
    ax.title.set_fontsize(16)

    xticks = range(0, len(df))
    xlabels = ["%i" % (q + 1) for q in xticks]
    plt.xlim(xticks[0] - 1, xticks[-1] + 1)
    plt.xticks(xticks, xlabels, rotation=0)

    plt.xlabel("Query")
    plt.ylabel("Running Time (ms)")

    for item in ([ax.xaxis.label, ax.yaxis.label] +
                 ax.get_xticklabels() + ax.get_yticklabels()):
        item.set_fontsize(12)

    dt = date.today().isoformat()
    nm = os.path.join("output/{}".format(method), "{}_{}_bar_{}_{}".format(kind, name, method, dt))
    
    normal_params = {"bbox_inches": "tight", "dpi": 1200}
    plot_list = [
        (".png", normal_params),
        ("_transparent.png", funcy.merge(normal_params, {"transparent": True})),
        (".eps", normal_params),
    ]
    for suffix, params in plot_list:
        plt.savefig(nm + suffix, **params)
    df.to_csv(nm + ".csv")
    
    return f


    

In [9]:

    

f = plot(times, "regression_tree")


    

    






<matplotlib.figure.Figure at 0x7f9c2c470668>






    

In [10]:

    

f = plot(times, "mean")


    

    






<matplotlib.figure.Figure at 0x7f9c16e685f8>






    

In [11]:

    

f = plot(times, "hot_deck")


    

    






<matplotlib.figure.Figure at 0x7f9c1f5e5208>