Initial implementation of Linear Regression functionalities on ml pac…

…kage path.

lib/zhvlib/ML.cc

@@ -0,0 +1,175 @@
+/*
+ * Copyright (c) 2024 - Nathanne Isip
+ * This file is part of Zhivo.
+ * 
+ * Zhivo is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published
+ * by the Free Software Foundation, either version 3 of the License,
+ * or (at your option) any later version.
+ * 
+ * Zhivo is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ * 
+ * You should have received a copy of the GNU General Public License
+ * along with Zhivo. If not, see <https://www.gnu.org/licenses/>.
+ */
+
+#include "zhvlib/ML.hpp"
+
+#include <cmath>
+#include <exception>
+#include <vector>
+
+static inline std::vector<double> arrayToDoubleVector(std::vector<DynamicObject> array) {
+    std::vector<double> values(array.size());
+
+    #pragma omp parallel for
+    for(size_t i = 0; i < array.size(); i++) {
+        if(!array[i].isNumber())
+            throw std::runtime_error("Value from array is not a number");
+
+        values[i] = array[i].getNumber();
+    }
+
+    return values;
+}
+
+static inline double calculateMean(std::vector<DynamicObject> array) {
+    std::vector<double> values = arrayToDoubleVector(array);
+    size_t arraySize = array.size();
+    double sum = 0.0;
+
+    #pragma omp parallel for reduction(+:sum)
+    for(size_t i = 0; i < arraySize; i++)
+        sum += values[i];
+
+    return sum / arraySize;
+}
+
+ZHIVO_FUNC(ml_trendline_calculate) {
+    if(args.size() != 2)
+        throw std::runtime_error(
+            "Expecting 2 argument, got " +
+                std::to_string(args.size()) +
+                "."
+        );
+
+    DynamicObject xObj = args.at(0),
+        yObj = args.at(1);
+
+    if(!xObj.isArray() || !yObj.isArray())
+        throw std::runtime_error("Parameter x and y must be both number array");
+
+    std::vector<DynamicObject> xObjArray = *xObj.getArray();
+    std::vector<DynamicObject> yObjArray = *yObj.getArray();
+
+    if(xObjArray.size() != yObjArray.size())
+        throw std::runtime_error("Data set size of x and y did not match");
+
+    double x = calculateMean(xObjArray),
+        y = calculateMean(yObjArray),
+        numerator = 0.0,
+        denominator = 0.0;
+
+    #pragma omp parallel for
+    for(size_t i = 0; i < xObjArray.size(); i++) {
+        numerator += (xObjArray[i].getNumber() - x) *
+            (yObjArray[i].getNumber() - y);
+
+        denominator += (xObjArray[i].getNumber() - x) *
+            (xObjArray[i].getNumber() - x);
+    }
+
+    double slope = numerator / denominator;
+    std::vector<DynamicObject> returnValues;
+
+    returnValues.emplace_back(DynamicObject(slope));
+    returnValues.emplace_back(DynamicObject(y - slope * x));
+
+    return DynamicObject(std::make_shared<std::vector<DynamicObject>>(
+        returnValues
+    ));
+}
+
+ZHIVO_FUNC(ml_trendline_calculateRmse) {
+    if(args.size() != 3)
+        throw std::runtime_error(
+            "Expecting 3 argument, got " +
+                std::to_string(args.size()) +
+                "."
+        );
+
+    DynamicObject xObj = args.at(0),
+        yObj = args.at(1),
+        model = args.at(2);
+    std::vector<DynamicObject> regModel = *model.getArray();
+
+    if(!model.isArray() || regModel.size() != 2)
+        throw std::runtime_error("Invalid linear regression model");
+
+    if(!xObj.isArray() || !yObj.isArray())
+        throw std::runtime_error("Parameter x and y must be both number array");
+
+    std::vector<DynamicObject> xObjArray = *xObj.getArray();
+    std::vector<DynamicObject> yObjArray = *yObj.getArray();
+
+    if(xObjArray.size() != yObjArray.size())
+        throw std::runtime_error("Data set size of x and y did not match");
+
+    double sumSquaredErrs = 0.0;
+    size_t paramSize = xObjArray.size();
+
+    #pragma omp parallel for
+    for(size_t i = 0; i < paramSize; i++) {
+        std::vector<DynamicObject> model;
+        model.emplace_back(regModel.at(0));
+        model.emplace_back(regModel.at(1));
+
+        std::vector<DynamicObject> params;
+        params.emplace_back(DynamicObject(
+            std::make_shared<std::vector<DynamicObject>>(model)
+        ));
+        params.emplace_back(xObjArray[i]);
+
+        double yPred = ml_trendline_predict(symtab, params)
+            .getNumber();
+        double error = yObjArray[i].getNumber() - yPred;
+
+        sumSquaredErrs += error * error;
+    }
+
+    return DynamicObject(std::sqrt(sumSquaredErrs / paramSize));
+}
+
+ZHIVO_FUNC(ml_trendline_predict) {
+    if(args.size() != 2)
+        throw std::runtime_error(
+            "Expecting 3 argument, got " +
+                std::to_string(args.size()) +
+                "."
+        );
+
+    DynamicObject model = args.at(0),
+        value = args.at(1);
+    std::vector<DynamicObject> regModel = *model.getArray();
+
+    if(!model.isArray() || regModel.size() != 2)
+        throw std::runtime_error("Invalid linear regression model");
+
+    if(!value.isNumber())
+        throw std::runtime_error("Cannot predict linear regression value for non-numbers");
+
+    DynamicObject slope = regModel.at(0),
+        intercept = regModel.at(1);
+
+    if(!slope.isNumber() || !intercept.isNumber())
+        throw std::runtime_error("Linear regression model's slope and intercept must be a number");
+
+    return DynamicObject(
+        slope.getNumber() *
+        value.getNumber() +
+        intercept.getNumber()
+    );
+}

lib/zhvlib/ML.hpp

@@ -0,0 +1,32 @@
+/*
+ * Copyright (c) 2024 - Nathanne Isip
+ * This file is part of Zhivo.
+ * 
+ * Zhivo is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published
+ * by the Free Software Foundation, either version 3 of the License,
+ * or (at your option) any later version.
+ * 
+ * Zhivo is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ * 
+ * You should have received a copy of the GNU General Public License
+ * along with Zhivo. If not, see <https://www.gnu.org/licenses/>.
+ */
+
+#ifndef ZHIVO_STDLIB_ML_CC
+#define ZHIVO_STDLIB_ML_CC
+
+#include "ZhivoLibrary.hpp"
+
+ZHIVO_LIB_START
+
+ZHIVO_FUNC(ml_trendline_calculate);
+ZHIVO_FUNC(ml_trendline_calculateRmse);
+ZHIVO_FUNC(ml_trendline_predict);
+
+ZHIVO_LIB_END
+
+#endif