Added lognormal test; Initialize with sample; Improved histogram; Closes

#1
open-risk · Feb 22, 2024 · 269cc49 · 269cc49
1 parent 01a8bbf
commit 269cc49
Show file tree

Hide file tree

Showing 15 changed files with 195 additions and 98 deletions.
diff --git a/README.md b/README.md
@@ -7,18 +7,17 @@ A C++ library for the calculation of various tail risk measures
 * Implemented [Risk Measures](RiskMeasures.md)
 * Mathematical Documentation available [here](https://www.openriskmanual.org/wiki/Category:Tail_Risk) 
 
-
 ## Dependencies
 
-* Eigen (Data container and calculation library)
-* Poco (For parsing JSON inputs)
+* Eigen (Data container and linear algebra calculation library)
+* Poco (For parsing JSON inputs and other utilities)
 * Stats (A C++ header-only library of statistical distribution functions.)
 
 ## Example
 The data directory contains sample datafiles with various sampled distributions
 
 ```c++
-    // Read in some data for a type 0 representation (discrete distribution)
+// Read in some data for a type 0 representation (discrete distribution)
 int LossGrid = 1000;
 int DataType = 0;
 RandomVar L(LossGrid, DataType);

diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
@@ -1,4 +1,4 @@
- ##  Copyright (C) 2020-2023 Open Risk (www.openriskmanagement.com)
+ ##  Copyright (C) 2020-2024 Open Risk (www.openriskmanagement.com)
  ##
  ##  This file is part of the tailRisk C++ library.
  ##
@@ -36,7 +36,7 @@ include_directories(..)
 include_directories(.)
 
 # tailRisk version
-set(VERSION 0.1.0)
+set(VERSION 0.2.0)
 
 set(SOURCE_FILES
         random_var.cpp
@@ -49,10 +49,10 @@ set(TEST_FILES
         ../testing/rnd_statistics/test_random_var_normal.cpp
         ../testing/rnd_statistics/test_random_var_uniform.cpp
         ../testing/rnd_statistics/test_random_var_exponential.cpp
-        ../testing/test_histogram.cpp
-        ../testing.cpp)
+        ../testing/rnd_statistics/test_random_var_lognormal.cpp
+        ../testing/test_histogram.cpp)
 
-find_package(Poco REQUIRED COMPONENTS Foundation Net NetSSL Util Data DataSQLite)
+find_package(Poco REQUIRED COMPONENTS Foundation Net Util Data)
 find_package(Catch2 REQUIRED)
 find_package(Eigen3 REQUIRED)
 find_package(statslib REQUIRED)
@@ -65,8 +65,9 @@ target_include_directories(tailRisk PRIVATE ${statslib_INCLUDE_DIRS})
 target_include_directories(tailRisk PRIVATE ${gcem_INCLUDE_DIRS})
 target_link_libraries(tailRisk
         Poco::Foundation
+        Poco::Data
+        Poco::JSON
         Poco::Net
-        Poco::NetSSL
         Poco::Util)
 
 # CATCH2 TESTING
@@ -81,11 +82,10 @@ target_include_directories(unit_testing PRIVATE ${Catch2_INCLUDE_DIRS})
 target_include_directories(unit_testing PRIVATE ${Eigen3_INCLUDE_DIRS})
 target_include_directories(unit_testing PRIVATE ${statslib_INCLUDE_DIRS})
 target_include_directories(unit_testing PRIVATE ${gcem_INCLUDE_DIRS})
-target_link_libraries(unit_testing
+target_link_libraries(unit_testing Catch2::Catch2WithMain
         Poco::Data
-        Poco::DataSQLite
         Poco::Foundation
         Poco::Net
-        Poco::NetSSL
-        Catch2::Catch2WithMain)
+        Poco::Util
+        Poco::JSON)
 
diff --git a/src/conanfile.txt b/src/conanfile.txt
@@ -5,6 +5,12 @@ statslib/3.2.0
 gcem/1.16.0
 catch2/3.4.0
 
+[options]
+poco*:enable_data_odbc=False
+poco*:enable_data_mysql=False
+poco*:enable_mongodb=False
+poco*:enable_redis=False
+
 [generators]
 CMakeDeps
 CMakeToolchain
diff --git a/src/main.cpp b/src/main.cpp
@@ -1,6 +1,6 @@
 /*################################################################################
   ##
-  ##   Copyright (C) 2020-2023 Open Risk (www.openriskmanagement.com)
+  ##   Copyright (C) 2020-2024 Open Risk (www.openriskmanagement.com)
   ##
   ##   This file is part of the tailRisk C++ library.
   ##
@@ -30,11 +30,23 @@
 
 int main(int argc, char *argv[]) {
 
+    // Example 1
+    int Grid = 1;
+    int DataType = 0;
+    RandomVar D(Grid, DataType);
+    std::string filename =  "../data/example1.json";
+    D.ReadFromJSON(filename);
+    D.Print();
+    D.Cumulative();
+    D.Probability();
+
+    // Example 5
     // Reading in some data for a type 0 representation (discrete distribution)
     int LossGrid = 1000;
-    int DataType = 0;
+    DataType = 0;
     RandomVar L(LossGrid, DataType);
-    std::string filename =  "../../data/example5.json";
+    // ATTN: Make sure this points to the correct path versus your executable!
+    filename =  "../data/example5.json";
     L.ReadFromJSON(filename);
     L.Print();
 
@@ -54,5 +66,13 @@ int main(int argc, char *argv[]) {
     std::cout << "Exceedance Probability: " << L.ExceedanceProbability(threshold) << std::endl;
     std::cout << "Mean Excess: " << L.MeanExcess(threshold ) << std::endl;
 
+    // Working with type 1 representations
+    int SampleSize = 10000;
+    DataType = 1;
+    RandomVar myR(SampleSize, DataType);
+    myR.Seed();
+    RandomVar H = myR.Histogram(10);
+    H.Print();
+
     return 0;
 }
diff --git a/src/random_var.cpp b/src/random_var.cpp
@@ -1,6 +1,6 @@
 /*################################################################################
  ##
- ##  Copyright (C) 2020-2023 Open Risk (www.openriskmanagement.com)
+ ##  Copyright (C) 2020-2024 Open Risk (www.openriskmanagement.com)
  ##
  ##  This file is part of the tailRisk C++ library.
  ##
@@ -30,6 +30,7 @@
 
 
 #include <Poco/JSON/Parser.h>
+#include "stats.hpp"
 #include "random_var.h"
 
 using namespace Poco;
@@ -57,7 +58,7 @@ RandomVar &RandomVar::operator=(const RandomVar &R) {
 };
 
 /**
- * Sort the sampling data
+ * Sort the sampling data if Type 1, else do nothing
  */
 void RandomVar::Sort() {
     if (m_type == 1) {
@@ -74,21 +75,27 @@ void RandomVar::Sort() {
  * Bins are assumed equal and represented by their mid-point values
  */
 RandomVar RandomVar::Histogram(int Bins) {
-    RandomVar H(Bins+1,0);
+    RandomVar H(Bins,0);
     if (m_type == 1) {
         std::sort(m_S.begin(), m_S.end());
         double min_value = m_S[0];
         double max_value = m_S[m_S.size()-1];
         double bin_width = (max_value - min_value)/ (double) Bins;
         double sample_p = 1.0 / m_size;
-        for (int i = 0; i < Bins + 1; i++) {
-            H.setX(i, min_value + bin_width / 2.0 +  (double) (i) * bin_width);
+        for (int i = 0; i < Bins; i++) {
+            double x = min_value + bin_width / 2.0 +  (double) (i) * bin_width;
+            H.setX(i, x);
             H.setP(i, 0.0);
+            H.setC(i, 0.0);
         }
         for (int j = 0; j < m_size; j++) {
             int observation_bin = (int) ( (m_S[j] - min_value) / bin_width );
+            if (observation_bin > Bins - 1) {
+                observation_bin = Bins - 1;
+            }
             H.addP(observation_bin, sample_p);
         }
+        H.Cumulative();
     }
     return H;
 }
@@ -361,3 +368,15 @@ void RandomVar::ReadFromJSON(std::string &filename) {
         this->setC(i, object->getValue<double>("cumulative"));
     }
 }
+
+/**
+ * Seed with a sampled distribution
+ */
+void RandomVar::Seed() {
+    if (m_type == 1) {
+        stats::rand_engine_t engine(12376);
+        for (int i = 0; i < this->m_size; i++) {
+            this->setS(i, stats::runif(0.0, 1.0, engine));
+        }
+    }
+}
diff --git a/src/random_var.h b/src/random_var.h
@@ -1,6 +1,6 @@
 /*################################################################################
   ##
-  ##   Copyright (C) 2020-20223 Open Risk (www.openriskmanagement.com)
+  ##   Copyright (C) 2020-2024 Open Risk (www.openriskmanagement.com)
   ##
   ##   This file is part of the tailRisk C++ library.
   ##
@@ -35,9 +35,9 @@ class RandomVar {
 
 public:
 
-    // constructor with size and empirical distribution type
-    // Type 0 -> histogram storage
-    // Type 1 -> sample storage
+    // Constructor with size and empirical distribution type
+    // Type 0 -> Histogram type Storage
+    // Type 1 -> Sample type Storage
 
     RandomVar(size_t S, int type) {
         if (type == 0) {
@@ -51,12 +51,11 @@ class RandomVar {
             m_S.resize(S);
             m_size = S;
         } else {
-            std::cout << "Error in random variable representation type" << std::endl;
+            std::cout << "Error in Random Variable representation type: Must be 0 or 1" << std::endl;
         }
     }
 
     // constructor directly from existing data
-    // TODO generalize data type to accommodate different accuracy requirements
 
     RandomVar(Eigen::ArrayXd x, Eigen::ArrayXd p, const int size) {
         m_type = 0;
@@ -86,6 +85,8 @@ class RandomVar {
         return m_type;
     };
 
+    // Getters and Setters
+
     [[nodiscard]] double getP(int index) const {
         return m_P[index];
     };
@@ -102,6 +103,30 @@ class RandomVar {
         return m_S[index];
     };
 
+    void setP(int index, double arg) {
+        m_P[index] = arg;
+    };
+
+    void setC(int index, double arg) {
+        m_C[index] = arg;
+    };
+
+    void setX(int index, double arg) {
+        m_X[index] = arg;
+    };
+
+    void setS(int index, double arg) {
+        m_S[index] = arg;
+    };
+
+    // Increment probability mass
+
+    void addP(int index, double arg) {
+        m_P[index] += arg;
+    };
+
+    // Statistical Measures
+
     [[nodiscard]] double Average();
 
     [[nodiscard]] double Mean();
@@ -130,25 +155,7 @@ class RandomVar {
 
     [[nodiscard]] int Quantile_Index(double alpha);
 
-    void setP(int index, double arg) {
-        m_P[index] = arg;
-    };
-
-    void setC(int index, double arg) {
-        m_C[index] = arg;
-    };
-
-    void addP(int index, double arg) {
-        m_P[index] += arg;
-    };
-
-    void setX(int index, double arg) {
-        m_X[index] = arg;
-    };
-
-    void setS(int index, double arg) {
-        m_S[index] = arg;
-    };
+    // Various Operations
 
     void Sort();
 
@@ -162,6 +169,8 @@ class RandomVar {
 
     void Print();
 
+    void Seed();
+
 private:
     // 0 Type: exact representation (discrete probabilities view)
     // 1 Type: sampling representation (distribution sampling view)

diff --git a/testing.cpp b/testing.cpp
diff --git a/testing/rnd_statistics/test_random_var_bernulli.cpp b/testing/rnd_statistics/test_random_var_bernulli.cpp
@@ -1,6 +1,6 @@
 /*################################################################################
   ##
-  ##   Copyright (C) 2020-2023 Open Risk (www.openriskmanagement.com)
+  ##   Copyright (C) 2020-2024 Open Risk (www.openriskmanagement.com)
   ##
   ##   This file is part of the tailRisk C++ library.
   ##
@@ -41,6 +41,11 @@ double bernulli_var(size_t number, double param) {
     return myR.Variance();
 }
 
+/*
+ * Numbers based on the Stats test
+ * https://github.com/kthohr/stats/blob/master/tests/rand/rbern.cpp
+*/
+
 double prob_par = 0.75;
 double m_bernulli_mean = prob_par;
 double m_bernulli_var = prob_par * (1.0 - prob_par);