Fixed error with distribute subsidies optimization.

Added new data files with firm configurations.

ace.py

@@ -5,6 +5,7 @@
 
 from create_firms import create_firms
 from match import match
+from distribute_funds import distribute_funding
 
 import pandas
 import random
@@ -27,7 +28,7 @@
 firm_configurations = ["firm_info_1.csv", "firm_info_1000_10_10.csv", "firm_info_5000_5000_5000.csv", "firm_info_10000_1000_100.csv",
                        "firm_info_10_10_10.csv", "firm_info_100_100_100.csv", "firm_info_10000_1000_100_10.csv", "firm_info_200000.csv"]
 
-regressions = ['loglinear','bayes', 'linear']
+regressions = ['bayes', 'linear']
 regression_types = ["average", "total"]
 
 distribute_subsidies = [True, False]
@@ -36,27 +37,33 @@
 
 with open("output.csv", "w", newline='') as output_file:
     writer = csv.DictWriter(output_file, delimiter=';',
-                            fieldnames=["seed", "firm_configuration", "regression", "regression_type", "distribute_subsidies",
+                            fieldnames=["seed", "firm_configuration", "regression_type", "regression", "distribute_subsidies",
                                         "disturb_result", "disturb_coefficients", "step", "mape", "r2_score"])
     writer.writeheader()
 output_file.close()
 
-for seed in range(1000):
-    for firm_config in firm_configurations:
+for firm_config in firm_configurations:
+    for seed in range(1000):
+        random.seed(seed)
         firm_info = create_firms(pandas.read_csv(firm_config, sep = ";", decimal = ","), history['employees'][0])
         firms = copy.deepcopy(firm_info)
         match_info = []
         for step in range(steps):
             if step != 0:
                 firms = match(firms, history['employees'][step] - history['employees'][step - 1])
             match_info.append(firms)
-        for regression in regressions:
-            for regression_type in regression_types:
-                for distribute_subsidy in distribute_subsidies:
+        for distribute_subsidy in distribute_subsidies:
+            if distribute_subsidy:
+                random.seed(seed)
+                distribute_subsidies_info = []
+                for step in range(steps):
+                    distribute_subsidies_info.append(distribute_funding(history, history['budget'][step], len(firm_info)))
+            for regression in regressions:
+                for regression_type in regression_types:
                     for disturb_result in disturb_results:
                         for disturb_coefficient in disturb_coefficients:
-                            Scenario("poland.csv", firm_info, match_info, firm_config, regression_type, distribute_subsidy, disturb_result,
-                                     disturb_coefficient, regression, seed).run(steps)
+                            Scenario("poland.csv", firm_info, match_info, distribute_subsidies_info, firm_config, regression_type,
+                                     distribute_subsidy, disturb_result, disturb_coefficient, regression, seed).run(steps)
 
 #scenario = Scenario("poland.csv", "firm_info_1000_10_10.csv", seed = 1)
 

distribute_funds.py

@@ -0,0 +1,15 @@
+import random
+
+def distribute_funding(history, subsidies, firm_number):
+    distributed_subsidies = []
+    subsidies_mean = history['subsidies'].mean()
+    subsidies_sd = history['subsidies'].std()
+    for i in range(firm_number):
+        new_subsidy = random.normalvariate(subsidies_mean, subsidies_sd)
+        new_subsidy = new_subsidy if new_subsidy > 0 else 0
+        distributed_subsidies.append(new_subsidy)
+    total = sum(distributed_subsidies)
+    total = total if total != 0 else 1
+    for i in range(firm_number):
+        distributed_subsidies[i] = distributed_subsidies[i] * subsidies / total
+    return distributed_subsidies

firm_info_10000_1000_100_10.csv

@@ -0,0 +1,5 @@
+number;workers;sd
+10000;5;1
+1000;25;10
+100;110;20
+10;1000;300

firm_info_1000_10_10.csv

@@ -0,0 +1,4 @@
+number;workers;sd
+1000;25;10
+10;110;20
+10;1000;300

firm_info_200000.csv

@@ -0,0 +1,4 @@
+number;workers;sd
+169162;25;10
+6156;110;20
+1510;1000;300

scenario.py

@@ -10,8 +10,9 @@
 import csv
 
 class Scenario():
-    def __init__(self, history, firm_info, match_info, firm_config, regression_type = "total", distribute_subsidies = False, disturb_result = False,
-                 disturb_coefficients = False, regression = "bayes", seed = 1, output_file = "output.csv",):
+    def __init__(self, history, firm_info, match_info, distribute_subsidies_info, firm_config, regression_type = "total",
+                 distribute_subsidies = False, disturb_result = False, disturb_coefficients = False, regression = "bayes",
+                 seed = 1, output_file = "output.csv",):
         random.seed(seed)
         self.seed = seed
         self.history = pandas.read_csv(history, sep = ";", decimal = ",")
@@ -24,11 +25,13 @@ def __init__(self, history, firm_info, match_info, firm_config, regression_type
         self.regression = regression
         self.distribute_subsidies = distribute_subsidies
         if self.regression_type == 'average':
-            self.model = World(self.history['employees'][0], self.firm_info, self.history[['workers', 'subsidies', 'sales']],
-                               distribute_subsidies, disturb_result, disturb_coefficients, regression, regression_type, match_info)
+            self.model = World(self.history['employees'][0], firm_info, self.history[['workers', 'subsidies', 'sales']],
+                               distribute_subsidies, disturb_result, disturb_coefficients, regression, regression_type, match_info,
+                               distribute_subsidies_info)
         else:
-            self.model = World(self.history['employees'][0], self.firm_info, self.history[['employees', 'budget', 'revenues']],
-                               distribute_subsidies, disturb_result, disturb_coefficients, regression, regression_type, match_info)
+            self.model = World(self.history['employees'][0], firm_info, self.history[['employees', 'budget', 'revenues']],
+                               distribute_subsidies, disturb_result, disturb_coefficients, regression, regression_type, match_info,
+                               distribute_subsidies_info)
 
         self.benchmark = self.history['revenues']
         self.workers_history = self.history['employees']

world.py

@@ -39,7 +39,7 @@ def other__init__(self, employees, firm_info, history, distribute_subsidies, dis
         self.t = 0
 
     def __init__(self, employees, firm_info, history, distribute_subsidies, disturb_result, disturb_coefficients,
-                     regression, regression_type, match_info):
+                     regression, regression_type, match_info, distribute_subsidies_info):
 
         self.distribute_subsidies = distribute_subsidies
         self.disturb_result = disturb_result
@@ -69,6 +69,7 @@ def __init__(self, employees, firm_info, history, distribute_subsidies, disturb_
         self.workers = []
         self.t = 0
         self.match_info = match_info
+        self.distribute_subsidies_info = distribute_subsidies_info
 
     def set_parameters(self):
         for firm in self.firms:
@@ -114,7 +115,8 @@ def step(self, subsidies, employees):
         sold = 0
         workers = 0
         if self.distribute_subsidies:
-            distributed_subsidies = self.distribute_funding(subsidies)
+            #distributed_subsidies = self.distribute_funding(subsidies)
+            distributed_subsidies = self.distribute_subsidies_info[self.t]
         else:
             distributed_subsidies = [subsidies/len(self.firms)] * len(self.firms)
         for i, firm in enumerate(self.firms):