source: branches/3022-FastFunctionExtraction/UnitTests/BasisFunctionUtilsTests.cs @ 17779

using System.Collections.Generic;
using System.Linq;
using HeuristicLab.Algorithms.DataAnalysis.FastFunctionExtraction;
using HeuristicLab.Data;
using HeuristicLab.Problems.DataAnalysis;
using Microsoft.VisualStudio.TestTools.UnitTesting;

namespace UnitTests {

    [TestClass]
    public class BasisFunctionUtilsTests {
        private static readonly HashSet<double> defaultExponents = new HashSet<double> { 0.5, 1.0, 2.0 };
        private static readonly HashSet<NonlinearOperator> defaultNonlinFuncs = new HashSet<NonlinearOperator> { NonlinearOperator.Abs, NonlinearOperator.Log, NonlinearOperator.Sin, NonlinearOperator.Cos };
        private static readonly Approach defaultApproach = new Approach(false, false, false, false, false, defaultExponents, defaultNonlinFuncs, 10, 1, 0.2, 0.8, 5);
        private static readonly IDataset defaultDataset = new Dataset(new string[] { "x1", "x2", "x3", "y" }, new double[,] {
            { 1, 1, -3, 7 },
            { 2, 3, -1, 12 },
            { 3, -3, 0, 0 },
            { 4, -2, 1, -8 },
            { 15, -8, 5, 2 }
        });

        [TestMethod]
        public void CreateBasisFunctions_ZeroFeatures_AllFalse() {
            var data = new RegressionProblemData(defaultDataset, new string[] { }, "y");
            var basisFunctions = BFUtils.CreateBasisFunctions(data, defaultApproach);
            Assert.AreEqual(basisFunctions.Count(), 0);
        }

        [TestMethod]
        public void CreateBasisFunctions_TwoFeatures_All_False() {
            var data = new RegressionProblemData(defaultDataset, new string[] { "x1", "x2" }, "y");
            var basisFunctions = BFUtils.CreateBasisFunctions(data, defaultApproach);
            Assert.AreEqual(basisFunctions.Count(), 2);
            var bf = basisFunctions.Last();
            Assert.AreEqual(bf, new SimpleBasisFunction("x2"));
        }

        [TestMethod]
        public void CreateBasisFunctions_TwoFeatures_AllowExponents() {
            var appr = defaultApproach;
            appr.AllowExp = true;
            appr.Exponents = defaultExponents;
            var data = new RegressionProblemData(defaultDataset, new string[] { "x1", "x2" }, "y");
            var basisFunctions = BFUtils.CreateBasisFunctions(data, appr);
            // -1 because you sqrt("x2") is not valid
            Assert.AreEqual(2 * appr.Exponents.Count - 1, basisFunctions.Count());
            var bf = basisFunctions.Last();
            Assert.AreEqual(bf, new SimpleBasisFunction("x2", exponent: defaultExponents.Last()));
        }

        [TestMethod]
        public void CreateSimpleBases_NoFeatures() {
            var data = new RegressionProblemData(defaultDataset, new string[] { }, "y");
            var bfs = BFUtils.CreateSimpleBases(data, new HashSet<double> { 1, 2, 0.5 }, new HashSet<NonlinearOperator> { NonlinearOperator.Abs }).ToArray();
            // num_features * num_exponents * num_nonlinearOperators
            Assert.AreEqual(0 * 3 * 1, bfs.Length);
        }

        [TestMethod]
        public void CreateSimpleBases_NoExponents() {
            var data = new RegressionProblemData(defaultDataset, new string[] { "x1" }, "y");
            var bfs = BFUtils.CreateSimpleBases(data, new HashSet<double> { }, new HashSet<NonlinearOperator> { NonlinearOperator.None }).ToArray();
            // num_features * num_exponents * num_nonlinearOperators
            Assert.AreEqual(1 * 0 * 1, bfs.Length);

            bfs = BFUtils.CreateSimpleBases(data, new HashSet<double> { 1 }, new HashSet<NonlinearOperator> { }).ToArray();
            Assert.AreEqual(1 * 1 * 0, bfs.Length);

            bfs = BFUtils.CreateSimpleBases(data, new HashSet<double> { 1 }, new HashSet<NonlinearOperator> { NonlinearOperator.None }).ToArray();
            Assert.AreEqual(1 * 1 * 1, bfs.Length);
            Assert.AreEqual(bfs.First(), new SimpleBasisFunction(feature: "x1", exponent: 1, op: NonlinearOperator.None, isNominator: true));

            bfs = BFUtils.CreateSimpleBases(data, new HashSet<double> { 2, 3 }, new HashSet<NonlinearOperator> { NonlinearOperator.None, NonlinearOperator.Log }).ToArray();
            Assert.AreEqual(1 * 2 * 2, bfs.Length);
            Assert.AreEqual(bfs.First(), new SimpleBasisFunction(feature: "x1", exponent: 2, op: NonlinearOperator.None, isNominator: true));
        }

        [TestMethod]
        public void CreateSimpleBases_NoNonlinearOperator() {
            var data = new RegressionProblemData(defaultDataset, new string[] { "x1", "x2" }, "y");
            var bfs = BFUtils.CreateSimpleBases(data, new HashSet<double> { 1, 2, 3 }, new HashSet<NonlinearOperator> { }).ToArray();
            // num_features * num_exponents * num_nonlinearOperators
            Assert.AreEqual(2 * 3 * 0, bfs.Length);
        }

        [TestMethod]
        public void CreateSimpleBases_ManyFeatures() {
            var data = new RegressionProblemData(defaultDataset, new string[] { "x1", "x2" }, "y");
            var bfs = BFUtils.CreateSimpleBases(data, new HashSet<double> { 1, 2, 3 }, new HashSet<NonlinearOperator> { NonlinearOperator.None }).ToArray();
            // num_features * num_exponents * (num_operators + 1)
            Assert.AreEqual(2 * 3 * 1, bfs.Length);
        }

        [TestMethod]
        public void CreateHingeBases() {
            var data = new RegressionProblemData(defaultDataset, new string[] { "x1", "x2" }, "y");
            var bfs = BFUtils.CreateSimpleBases(data, new HashSet<double> { 1, 2 }, new HashSet<NonlinearOperator> { NonlinearOperator.None, NonlinearOperator.Sin }).ToArray();

            var numSimpleBasisFuncs = bfs.Length;
            int num_thrs = 2;
            var hingeBases = BFUtils.CreateHingeBases(data, bfs, 0, 1, num_thrs).ToArray();
            Assert.AreEqual(2 * numSimpleBasisFuncs * num_thrs, hingeBases.Length);
        }

        [TestMethod]
        public void CreateHingeBases_CheckOtherBFValues() {
            var data = new RegressionProblemData(defaultDataset, new string[] { "x3" }, "y");
            var bfs = BFUtils.CreateSimpleBases(data, new HashSet<double> { 1 }, new HashSet<NonlinearOperator> { NonlinearOperator.None }).ToArray();
            Assert.AreEqual(1, bfs.Length);

            int num_thrs = 2;
            var partition = new IntRange(0, data.Dataset.Rows);
            var hingeBases = BFUtils.CreateHingeBases(data, bfs, 0, 1, num_thrs, partition).ToArray();
            Assert.AreEqual(2 * 1 * num_thrs, hingeBases.Length);
            foreach(var hingeBase in hingeBases) {
                Assert.AreEqual(hingeBase.Feature, "x3");
                Assert.AreEqual(hingeBase.Exponent, 1);
                Assert.AreEqual(hingeBase.IsDenominator, true);
            }
        }

        [TestMethod]
        public void CreateHingeBases_CheckArbitraryStartThrAndEndThr() {
            var data = new RegressionProblemData(defaultDataset, new string[] { "x3" }, "y");
            var bfs = BFUtils.CreateSimpleBases(data, new HashSet<double> { 1 }, new HashSet<NonlinearOperator> { NonlinearOperator.None }).ToArray();
            Assert.AreEqual(1, bfs.Length);

            int num_thrs = 2;
            var partition = new IntRange(0, data.Dataset.Rows);
            var hingeBases = BFUtils.CreateHingeBases(data, bfs, 0.5, 0.75, num_thrs, partition).ToArray();
            Assert.AreEqual(2 * num_thrs, hingeBases.Length);
            Assert.AreEqual(hingeBases.Count(hingeBase => hingeBase.Threshold == 1), 2);
            Assert.AreEqual(hingeBases.Count(hingeBase => hingeBase.Threshold == 3), 2);
        }

        [TestMethod]
        public void CreateHingeBases_CheckManyThrs() {
            var data = new RegressionProblemData(defaultDataset, new string[] { "x3" }, "y");
            var bfs = BFUtils.CreateSimpleBases(data, new HashSet<double> { 1 }, new HashSet<NonlinearOperator> { NonlinearOperator.None }).ToArray();
            Assert.AreEqual(1, bfs.Length);

            int num_thrs = 9;
            var partition = new IntRange(0, data.Dataset.Rows);

            // i.e. from -11 to 13
            double start_thr = -1;
            double end_thr = 2;

            var hingeBases = BFUtils.CreateHingeBases(data, bfs, start_thr, end_thr, num_thrs, partition).ToArray();
            Assert.AreEqual(2 * num_thrs, hingeBases.Length);
            Assert.AreEqual(hingeBases.Count(hingeBase => hingeBase.Threshold == -11), 2);
            Assert.AreEqual(hingeBases.Count(hingeBase => hingeBase.Threshold == -8), 2);
            Assert.AreEqual(hingeBases.Count(hingeBase => hingeBase.Threshold == 13), 2);
        }
    }
}