summary pruning data analysis

projects/summary-pruning/index.org

+#+TITLE: Summary Pruning
+#+PROPERTY: header-args :session main :exports both :results output
+
+* DBLP 
+#+BEGIN_src jupyter-python :exports none :file dblp-test.png
+  import pandas as pd
+  import numpy as np
+  import matplotlib.patches as mpatches
+  import matplotlib.pyplot as plt
+
+  TIMEOUT = 10 * 60 * 1000 # 10 min in millisec
+
+  def load_csv(path, index=None, columns=None):
+        if path is None:
+              return pd.DataFrame([]).reindex(index=index, columns=columns)
+        df = pd.read_csv(path, sep='\t').groupby(['INPUT']).mean()
+        if index is None:
+              return df
+        else:
+              return df.reindex(index=index)
+
+  def add_bars(axs, x_pos, df, width, color, time_label):
+        missed_index = df['TOTAL'].isnull()
+        unmissed_index = df['TOTAL'].notnull()
+        axs['time'].bar(x_pos[missed_index], TIMEOUT, width, fill=False, edgecolor='tomato', hatch="////")
+        axs['time'].bar(x_pos[unmissed_index], df['TOTAL'][unmissed_index], width, label=time_label, color=color)
+        axs['answers'].bar(x_pos, df['NANS'], width, label=time_label, color=color)
+        axs['ref'].bar(x_pos, df['NPROD_REF'], width, label="explored ref.", fill=False, edgecolor=color)
+        axs['ref'].bar(x_pos, df['POST_NPRUN'], width, label="pruned ref.", fill=False, hatch='////', edgecolor=color)
+        axs['ref'].bar(x_pos, df['POST_NCOV'], width, label="evaluated ref.", fill=False, hatch='\\\\\\\\', edgecolor=color)
+
+        axs['time_per'].bar(x_pos, df['TREF']/df['TOTAL'], width, label=time_label, color=color)
+        axs['time_per'].bar(x_pos, df['POST_TPRUN']/df['TOTAL'], width, label=time_label, edgecolor=color, fill=False, bottom=df['TREF']/df['TOTAL'], hatch='\\\\\\\\////')
+        axs['time_per'].bar(x_pos, df['POST_TCOV']/df['TOTAL'], width, label=time_label, edgecolor=color, fill=False, bottom=(df['TREF'] + df['POST_TPRUN'])/df['TOTAL'], hatch='..')
+        axs['time_per'].bar(x_pos, df['EXEC_TIME']/df['TOTAL'], width, label=time_label, fill=False, edgecolor=color, bottom=(df['TREF'] + df['POST_TPRUN'] + df['POST_TCOV'])/df['TOTAL'])
+
+
+  def plot(pre_path, post_path=None, emptyprun_path=None, ref_path=None, title=''):
+
+      df_pre = load_csv(pre_path)
+      index = df_pre.index
+      columns = df_pre.columns
+      df_post = load_csv(post_path, index, columns)
+      df_emptyprun = load_csv(emptyprun_path, index, columns)
+      df_ref = load_csv(ref_path, index, columns)
+
+      #-- Create and adjust new figure
+      fig = plt.figure(
+          constrained_layout=True,
+          figsize=(18,11)
+      )
+
+      # set_dpi(120)
+      # set_font_size(15)
+
+      #-- use the mosaic layout for custom sizes
+      axs = fig.subplot_mosaic(
+            [['time'],['time_per'],['ref'],['answers']],  #layout and axes handles. 
+            gridspec_kw={'height_ratios':[1, 1, 1, 1]},  #ratio between top plot and bottom plot
+            sharex=True,
+      )
+
+      #-- x ticks
+      X = np.arange(len(index))
+      axs['time'].set_xticks(X, index)
+      axs['ref'].set_xticks(X, index)
+      #-- plot data
+      width = 0.2
+      palette = plt.cm.tab20c
+      add_bars(axs, X-1.5*width, df_emptyprun, width, palette(14), "NO_PRUN")
+      add_bars(axs, X-.5*width, df_emptyprun, width, palette(8), "EMPTY_PRUN")
+      add_bars(axs, X+.5*width, df_post, width, palette(6), "POST")
+      add_bars(axs, X+1.5*width, df_pre, width, palette(0), "PRE")
+
+      #-- log scalen
+      axs['time'].set_yscale('log')
+      axs['ref'].set_yscale('log')
+      axs['answers'].set_yscale('log')
+
+      #-- set axis labels
+      axs['time'].set_ylabel('Time (ms)')
+      axs['time_per'].set_ylabel('part of time for each step')
+      axs['ref'].set_ylabel("Number of reformulation")
+      axs['answers'].set_ylabel('Number of answers')
+
+      #-- set grids
+      axs['time'].xaxis.grid()
+      axs['time_per'].xaxis.grid()
+      axs['ref'].xaxis.grid()
+      axs['answers'].xaxis.grid()
+
+      #-- set subplot titles
+      axs['time'].set_title(title)
+
+      #-- set subplot legends
+      #time_legend = axs['time'].legend()
+      thandles, tlabels = axs['time'].get_legend_handles_labels()
+      thandles.append(mpatches.Patch(fill=None, edgecolor='tomato', hatch='////'))
+      tlabels.append("Timeout/error")
+      axs['time'].legend(handles = thandles, labels = tlabels)
+
+      explored_ref = mpatches.Patch(fill=None,label='Explored ref.')
+      returned_ref = mpatches.Patch(fill=None,label='Returned ref.', hatch='////')
+      evaluated_ref = mpatches.Patch(fill=None,label='Evaluated ref.', hatch='\\\\\\\\')
+      axs['ref'].legend(handles=[explored_ref, returned_ref, evaluated_ref])
+
+      ref_time= mpatches.Patch(color='black', label='Reformulation')
+      prun_time= mpatches.Patch(fill=None,label='Post pruning', hatch='\\\\\\\\////')
+      cover_time = mpatches.Patch(fill=None, label='Cover', hatch='..')
+      ans_time = mpatches.Patch(fill=None,label='Evaluation')
+      axs['time_per'].legend(handles=[ans_time, cover_time, prun_time, ref_time])
+
+  plot('results/dblp-test-pre/stats.csv', 'results/dblp-test-post/stats.csv', 'results/dblp-test/stats.csv', 'results/dblp-test-noemptyp/stats.csv', "DBLP-test")
+#+END_src
+
+#+RESULTS:
+[[file:dblp-test.png]]
+
+** DBLP (SSWS)
+
+*Queries* : [[file:results/dblpmsc-pre/inputs/dblp.queries][dblp.queries]]
+
+#+BEGIN_src jupyter-python :file dblp.png
+  plot('results/dblpmsc-pre/stats.csv', 'results/dblpmsc-post/stats.csv', 'results/dblpmsc-ref/stats.csv', 'results/dblpmsc-ref0/stats.csv', 'DBLP')
+#+END_src
+
+#+RESULTS:
+[[file:dblp.png]]
+
+** DBLP (SSWS with constants)
+
+*Queries* : [[file:results/dblp-test-pre/inputs/dblp-test.queries][dblp-test.queries]]
+
+#+BEGIN_src jupyter-python :file dblp-test.png
+plot('results/dblp-test-pre/stats.csv', 'results/dblp-test-post/stats.csv', 'results/dblp-test/stats.csv', 'results/dblp-test-noemptyp/stats.csv', "DBLP with constant")
+#+END_src
+
+#+RESULTS:
+[[file:dblp-test.png]]
+
+* LUBM (on going)
+
+*Dataset* : LUBM100M
+*Queries* : [[file:results/lubm100m/inputs/lubm-damian.queries][lubm-damian.queries]]
+*Ontosql version* : ...
+
+#+BEGIN_src jupyter-python :file lubm.png 
+  plot('results/lubm100m/stats.csv', 'results/lubm50m/stats.csv')
+#+END_src
+
+#+RESULTS:
+[[file:lubm.png]]

publish.el

@@ -98,10 +98,11 @@ escape some problematic character"
       ;; ` character is very important
       ;; replace it by ' and functions inside object will not be executed
       ;; function call have to be prefixed with ,
-      `(,@(uca-website-create-project-configuration
-           "dev"
-           "~/work/.public_html"
-           "")
+              `(
+           ;;      ,@(uca-website-create-project-configuration
+           ;; "dev"
+           ;; "~/work/.public_html"
+           ;; "")
         ,@(uca-website-create-project-configuration
            "prod"
            "/scp:uca:~/public_html/"