updates the automation scripts

plot.py

@@ -11,18 +11,18 @@
 # List of indexes
 difficulty = ['8-12', '12-16', '16-20', '20-24', '24-28']
 # List of columns
-weapons = ['Barehand', 'Sword', 'Bow', 'Bomb Thrower', 'Shield', 'Cure Spell']
+weapon = ['Barehand', 'Sword', 'Bow', 'Bomb Thrower', 'Shield', 'Cure Spell']
 
 
 # Convert the array into a map
 def to_map(array, attribute):
-  shape = (len(difficulty), (len(weapons)))
+  shape = (len(difficulty), len(weapon))
   map = np.zeros(shape)
   i = 0
   for b in range(len(difficulty)):
-    for w in range(len(weapons)):
+    for w in range(len(weapon)):
       if array[i] is None:
-        map[b, w] = 0
+        map[b, w] = None
       else:
         map[b, w] = array[i][attribute]
       i += 1
@@ -31,7 +31,7 @@ def to_map(array, attribute):
 
 # Plot the heatmap
 def plot_heatmap(map, folder, filename, pop, max):
-  df = DataFrame(map, index=difficulty, columns=weapons)
+  df = DataFrame(map, index=difficulty, columns=weapon)
   color = sb.color_palette('viridis_r', as_cmap=True)
   ax = sb.heatmap(df, vmin=0, vmax=max, annot=True, cmap=color)
   ax.invert_yaxis()
@@ -46,7 +46,7 @@ def plot_charts(data, filename, folder):
   # Parse file
   obj = json.loads(data)
 
-  # Create the folder for the given parameters
+  # Create the folder for the entered parameters
   filename = filename.replace('.json', '')
   folder = folder +  os.path.sep + filename
   if os.path.isdir(folder):
@@ -64,44 +64,10 @@ def plot_charts(data, filename, folder):
   plot_heatmap(f_intermediate, folder, filename, 'fitness_intermediate', max)
   plot_heatmap(f_final, folder, filename, 'fitness_final', max)
 
-  # Print measurements
-  measure = {}
 
-  # Calculate mean fitness
-  measure['mean_f_initial'] = np.mean(f_initial)
-  measure['mean_f_intermediate'] = np.mean(f_intermediate)
-  measure['mean_f_final'] = np.mean(f_final)
 
-  # Calculate max fitness
-  measure['max_f_initial'] = np.max(f_initial)
-  measure['max_f_intermediate'] = np.max(f_intermediate)
-  measure['max_f_final'] = np.max(f_final)
-
-  # Calculate min fitness
-  measure['min_f_initial'] = np.min(f_initial)
-  measure['min_f_intermediate'] = np.min(f_intermediate)
-  measure['min_f_final'] = np.min(f_final)
-
-  # Calculate the standard deviation
-  measure['std_f_initial'] = np.std(f_initial)
-  measure['std_f_intermediate'] = np.std(f_intermediate)
-  measure['std_f_final'] = np.std(f_final)
-
-  # Serializing json
-  json_object = json.dumps(measure, indent = 4)
-
-  # Write JSON file
-  filename = folder + os.path.sep + filename + '-measure.json'
-
-  # Writing to sample.json
-  with open(filename, "w") as outfile:
-    outfile.write(json_object)
-
-
-# Target folder to save the charts
+# Create the folder for the charts
 CHART_FOLDER = 'charts'
-
-# Check if the folder for charts exists and create it if it does not exist
 if not os.path.isdir(CHART_FOLDER):
   os.mkdir(CHART_FOLDER)
 
@@ -115,7 +81,7 @@ def plot_charts(data, filename, folder):
 param_folder += sys.argv[5]       # Number of competitors
 
 
-# Read all the JSON files for the given parameters
+# Read all the JSON files for the entered parameters
 files = []
 filenames = []
 for p in Path(folder + param_folder).glob('*.json'):
@@ -124,7 +90,7 @@ def plot_charts(data, filename, folder):
     filenames.append(p.name)
 
 
-# Create the folder for the given parameters
+# Create the folder for the entered parameters
 folder = CHART_FOLDER + os.path.sep + param_folder
 if not os.path.isdir(folder):
   os.mkdir(folder)

run.py

@@ -1,4 +1,5 @@
 import os
+import sys
 import platform
 import random
 import numpy as np
@@ -8,10 +9,13 @@
 # --- Initialization
 
 # Initialize the random seed
-random.seed(0)
+seed = random.randrange(sys.maxsize)
+random.seed(seed)
+print("Seed:", seed)
+# random.seed(0)
 
 # Define the number of executions of each set of parameters
-executions = range(1)
+executions = range(3)
 
 
 
@@ -24,10 +28,10 @@
 populations = [25]
 
 # Mutation rates
-mutations = [30]
+mutations = [10]
 
-# Crossover rates
-crossovers = [80]
+# Mutation rates for genes
+genemutations = [30]
 
 # Competitors
 competitors = [3]
@@ -40,7 +44,7 @@
 if platform.system() == 'Linux':
   executable = './bin/Debug/net5.0/publish/EnemyGenerator '
 elif platform.system() == 'Windows':
-  executable = 'start bin\\Debug\\net5.0\\publish\\EnemyGenerator '
+  executable = 'bin\\Debug\\net5.0\\publish\\EnemyGenerator '
 else:
   print('This script is not able to run in this OS.')
   exit()
@@ -50,32 +54,23 @@
 os.system('dotnet publish')
 
 
-def run(ge, po, mu, cr, co):
+def run(ge, po, mu, gm, co):
   # Generate a random seed
   rs = random.randint(0, np.iinfo(np.int32).max - 1)
   # Build the parameters
   parameters = ""
-  for i in [rs, ge, po, mu, cr, co]:
+  for i in [rs, ge, po, mu, gm, co]:
     parameters += str(i) + ' '
   # Print parameters
   print('Parameters=[', parameters, ']')
   # Run algoritm for the current set of parameters
   os.system(executable + parameters)
 
-
-def plot(ge, po, mu, cr, co):
-  # Plot the charts for the current set of parameters
-  parameters = ''
-  for i in [ge, po, mu, cr, co]:
-    parameters += str(i) + ' '
-  os.system('python plot.py ' + parameters)
-
-
 # Variables to control the experiment progress
 total = len(generations) * \
   len(populations) * \
   len(mutations) * \
-  len(crossovers) * \
+  len(genemutations) * \
   len(competitors) * \
   len(executions)
 i = 1
@@ -85,31 +80,41 @@ def plot(ge, po, mu, cr, co):
 for ge in generations:
   for po in populations:
     for mu in mutations:
-      for cr in crossovers:
+      for gm in genemutations:
         for co in competitors:
           for e in executions:
             # Run execuble
-            run(ge, po, mu, cr, co)
+            run(ge, po, mu, gm, co)
             # Print progress
             print("%.2f" % ((i / total) * 100))
             i += 1
 
 
-# Variables to control the experiment progress
+# --- Plot charts of the experiment results
+
+def plot(ge, po, mu, gm, co):
+  parameters = ''
+  for i in [ge, po, mu, gm, co]:
+    parameters += str(i) + ' '
+  os.system('python plot.py ' + parameters)
+
+# Variables to control the plotting progress
 total = len(generations) * \
   len(populations) * \
   len(mutations) * \
-  len(crossovers) * \
+  len(genemutations) * \
   len(competitors)
 i = 1
 
-# Plot all the results
+# Plot charts for all sets of parameters
 print('Plotting')
 for ge in generations:
   for po in populations:
     for mu in mutations:
-      for cr in crossovers:
+      for gm in genemutations:
         for co in competitors:
+          # Plot charts
+          plot(ge, po, mu, gm, co)
           # Print progress
           print("%.2f" % ((i / total) * 100))
-          plot(ge, po, mu, cr, co)
+          i += 1

stats.py

@@ -0,0 +1,91 @@
+import os
+from pathlib import Path
+import json
+import numpy as np
+from pandas import DataFrame
+
+
+# List of indexes
+difficulty = ['8-12', '12-16', '16-20', '20-24', '24-28']
+# List of columns
+weapon = ['Barehand', 'Sword', 'Bow', 'Bomb Thrower', 'Shield', 'Cure Spell']
+
+
+# Convert the list of files to a map
+def to_map(array, attribute):
+  shape = (len(difficulty), len(weapon))
+  map = np.zeros(shape)
+  i = 0
+  for b in range(len(difficulty)):
+    for w in range(len(weapon)):
+      if array[i] is None:
+        map[b, w] = None
+      else:
+        map[b, w] = array[i][attribute]
+      i += 1
+  return map
+
+shape = (len(difficulty), len(weapon))
+mean_map = np.zeros(shape)
+std_map = np.zeros(shape)
+
+maps = []
+
+path = 'results' + os.path.sep + '100-25-30-80-3' + os.path.sep
+for p in Path(path).glob('*.json'):
+  with p.open() as f:
+    # Convert the read files into a map of fitness
+    # and add them to the list of maps
+    obj = json.loads(f.read())
+    maps.append(to_map(obj['final'], 'fitness'))
+
+for map in maps:
+  for l in range(len(difficulty)):
+    for e in range(len(weapon)):
+      print('%3.2f' % mean_map[l, e], end=' ')
+      if not np.isnan(map[l, e]):
+        mean_map[l, e] += map[l, e]
+    print()
+  print()
+
+mean_map = mean_map / 10
+
+# Uncomment to debug the mean map
+# for l in range(len(difficulty)):
+#   for e in range(len(weapon)):
+#     print('%3.2f' % mean_map[l, e], end=' ')
+#   print()
+
+for map in maps:
+  for l in range(len(difficulty)):
+    for e in range(len(weapon)):
+      if not np.isnan(map[l, e]):
+        std_map[l, e] += pow(map[l, e] - mean_map[l, e], 2)
+std_map = std_map / 10
+std_map = np.sqrt(std_map)
+
+# Uncomment to debug the std map
+# for l in range(len(difficulty)):
+#   for e in range(len(weapon)):
+#     print('%3.2f' % std_map[l, e], end=' ')
+#   print()
+
+# Merge the mean and std maps
+fmap = [ [ '' for e in range(len(weapon)) ] for l in range(len(difficulty)) ]
+for l in range(len(difficulty)):
+  for e in range(len(weapon)):
+    fmap[l][e] = '{:.2f}+-{:.2f}'.format(mean_map[l, e], std_map[l, e])
+
+# Uncomment to debug the merged map
+# for l in range(len(difficulty)):
+#   for e in range(len(weapon)):
+#     print(fmap[l][e], end=' ')
+#   print()
+
+# Print the resulting table
+df = DataFrame(fmap, index=difficulty, columns=weapon)
+print(df)
+
+# Uncomment to write a CSV file with the resulting table
+# filename = 'std_atual.csv'
+# df.to_csv(filename)