1.14.28.0: Add code for computing beta metric

hafniz · Feb 27, 2020 · 9d75d4f · 9d75d4f
1 parent de5f17e
commit 9d75d4f
Show file tree

Hide file tree

Showing 3 changed files with 41 additions and 13 deletions.
diff --git a/MLCore/Algorithm/KNNContext.cs b/MLCore/Algorithm/KNNContext.cs
@@ -103,5 +103,24 @@ public double GetAlphaValue(Instance testingInstance)
             }
             return alphas;
         }
+
+        public double GetBetaValue(Instance testingInstance)
+        {
+            int homoCount = TrainingInstances.Count(i => i.LabelValue == testingInstance.LabelValue);
+            IEnumerable<Instance> neighbors = GetNeighbors(testingInstance, homoCount - 1);
+            double c = neighbors.Where(i => i.LabelValue == testingInstance.LabelValue).Sum(i => 1.0 / EuclideanDistance(i, testingInstance));
+            double d = TrainingInstances.Where(i => i != testingInstance).Sum(i => 1.0 / EuclideanDistance(i, testingInstance));
+            return c / d;
+        }
+
+        public List<(Instance, double)> GetAllBetaValues()
+        {
+            List<(Instance, double)> betas = new List<(Instance, double)>();
+            foreach (Instance trainingInstance in TrainingInstances)
+            {
+                betas.Add((trainingInstance, GetBetaValue(trainingInstance)));
+            }
+            return betas;
+        }
     }
 }
diff --git a/MLCore/MLCore.csproj b/MLCore/MLCore.csproj
@@ -7,7 +7,7 @@
     <Nullable>enable</Nullable>
     <StartupObject>MLCore.Program</StartupObject>
     <ApplicationIcon>TextTemplate.ico</ApplicationIcon>
-    <Version>1.14.26.0</Version>
+    <Version>1.14.28.0</Version>
   </PropertyGroup>
 
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|AnyCPU'">

diff --git a/MLCore/Program.cs b/MLCore/Program.cs
@@ -1,10 +1,10 @@
-#define NOOP
+#define BETA_EXPR
 using System;
 using System.Collections.Generic;
-using System.ComponentModel.DataAnnotations;
 using System.IO;
 using System.Linq;
 using System.Text;
+using System.Threading.Tasks;
 using MLCore.Algorithm;
 
 namespace MLCore
@@ -519,19 +519,28 @@ static void Main()
 #endif
         #endregion
 
-        static void Main()
+        #region BETA_EXPR
+#if BETA_EXPR
+        public static int finishedCount = 0;
+
+        public static void Main()
+        {
+            Parallel.ForEach(Directory.EnumerateFiles("C:\\Users\\CHENH\\source\\repos\\MachineLearning\\Dataset\\A270\\original\\a270-raw"), filename => CalcBeta(filename));
+            Parallel.ForEach(Directory.EnumerateFiles("C:\\Users\\CHENH\\source\\repos\\MachineLearning\\Dataset\\UCI\\ECOC8030\\ECOC8030-raw"), filename => CalcBeta(filename));
+        }
+
+        public static void CalcBeta(string filename)
         {
-            int i = 1;
-            StringBuilder sb = new StringBuilder("filename,majority%\r\n");
-            foreach (string filename in Directory.EnumerateFiles("C:\\Users\\CHENH\\source\\repos\\MachineLearning\\Dataset\\UCI\\ECOC8030\\ECOC8030-raw"))
+            List<Instance> instances = CSV.ReadFromCsv(filename, null);
+            StringBuilder sb = new StringBuilder($"{string.Join(',', instances.First().Features.Select(f => f.Name))},label,beta\r\n");
+            foreach ((Instance instance, double beta) in new KNNContext(instances).GetAllBetaValues())
             {
-                List<string> labels = CSV.ReadFromCsv(filename, true).SelectColumn(^1);
-                int majorityCount = Math.Max(labels.Count(s => s == "0.0"), labels.Count(s => s == "1.0"));
-                sb.AppendLine($"{Path.GetFileNameWithoutExtension(filename)},{ majorityCount / (double)labels.Count}");
-                Console.WriteLine(i++);
+                sb.AppendLine($"{instance.Serialize()},{beta}");
             }
-            using StreamWriter sw = new StreamWriter("C:\\Users\\CHENH\\Desktop\\CBstats.csv");
-            sw.Write(sb);
+            File.WriteAllText($"C:\\Users\\CHENH\\Desktop\\beta\\{Path.GetFileName(filename)}", sb.ToString());
+            Console.WriteLine(++finishedCount);
         }
+#endif
+        #endregion
     }
 }