Added class for continuous valued stochastic processes

dynaml-core/src/main/scala-2.11/io/github/mandar2812/dynaml/models/StochasticProcess.scala

@@ -53,7 +53,7 @@ trait StochasticProcess[T, I, Y, W] extends Model[T, I, Y] {
   * @tparam W Implementing class of the posterior distribution
   *
   * */
-abstract class SecondOrderProcess[T, I, Y, K, M, W] extends StochasticProcess[T, I, Y, W] {
+trait SecondOrderProcess[T, I, Y, K, M, W] extends StochasticProcess[T, I, Y, W] {
 
   private val logger = Logger.getLogger(this.getClass)
 
@@ -72,3 +72,47 @@ abstract class SecondOrderProcess[T, I, Y, K, M, W] extends StochasticProcess[T,
 
 
 }
+
+/**
+  * @author mandar2812 date: 11/10/2016
+  *
+  * Blueprint for a continuous valued stochastic process, abstracts away the behavior
+  * common to sub-classes such as [[io.github.mandar2812.dynaml.models.gp.GPRegression]],
+  * [[io.github.mandar2812.dynaml.models.stp.StudentTRegression]] and others.
+  *
+  */
+abstract class ContinuousProcess[T, I, Y, W] extends StochasticProcess[T, I, Y, W] {
+
+  private val logger = Logger.getLogger(this.getClass)
+
+  /**
+    * Represents how many intervals away from the mean the error bars lie, defaults to 1
+    */
+  private var errorSigma: Int = 1
+
+  def _errorSigma = errorSigma
+
+  def errorSigma_(s: Int) = errorSigma = s
+
+  /**
+    * Draw three predictions from the posterior predictive distribution
+    * 1) Mean or MAP estimate Y
+    * 2) Y- : The lower error bar estimate (mean - sigma*stdDeviation)
+    * 3) Y+ : The upper error bar. (mean + sigma*stdDeviation)
+    **/
+  def predictionWithErrorBars[U <: Seq[I]](testData: U, sigma: Int): Seq[(I, Y, Y, Y)]
+
+  /**
+    * Returns a prediction with error bars for a test set of indexes and labels.
+    * (Index, Actual Value, Prediction, Lower Bar, Higher Bar)
+    * */
+  def test(testData: T): Seq[(I, Y, Y, Y, Y)] = {
+    logger.info("Generating predictions for test set")
+    //Calculate the posterior predictive distribution for the test points.
+    val predictionWithError = this.predictionWithErrorBars(dataAsIndexSeq(testData), errorSigma)
+    //Collate the test data with the predictions and error bars
+    dataAsSeq(testData).zip(predictionWithError).map(i => (i._1._1, i._1._2,
+      i._2._2, i._2._3, i._2._4))
+  }
+
+}

dynaml-core/src/main/scala-2.11/io/github/mandar2812/dynaml/models/gp/AbstractGPRegressionModel.scala

@@ -21,7 +21,7 @@ package io.github.mandar2812.dynaml.models.gp
 import breeze.linalg._
 import breeze.numerics.log
 import io.github.mandar2812.dynaml.kernels.{DiracKernel, LocalScalarKernel}
-import io.github.mandar2812.dynaml.models.SecondOrderProcess
+import io.github.mandar2812.dynaml.models.{ContinuousProcess, SecondOrderProcess}
 import io.github.mandar2812.dynaml.optimization.GloballyOptWithGrad
 import io.github.mandar2812.dynaml.probability.MultGaussianRV
 import org.apache.log4j.Logger
@@ -40,12 +40,11 @@ import org.apache.log4j.Logger
   *
   */
 abstract class AbstractGPRegressionModel[T, I](
-  cov: LocalScalarKernel[I],
-  n: LocalScalarKernel[I],
-  data: T, num: Int) extends
-  SecondOrderProcess[T, I, Double, Double, DenseMatrix[Double],
-  MultGaussianRV]
-with GloballyOptWithGrad {
+  cov: LocalScalarKernel[I], n: LocalScalarKernel[I],
+  data: T, num: Int)
+  extends ContinuousProcess[T, I, Double, MultGaussianRV]
+  with SecondOrderProcess[T, I, Double, Double, DenseMatrix[Double], MultGaussianRV]
+  with GloballyOptWithGrad {
 
   private val logger = Logger.getLogger(this.getClass)
 
@@ -223,18 +222,6 @@ with GloballyOptWithGrad {
 
   override def predict(point: I): Double = predictionWithErrorBars(Seq(point), 1).head._2
 
-  /**
-    * Returns a prediction with error bars for a test set of indexes and labels.
-    * (Index, Actual Value, Prediction, Lower Bar, Higher Bar)
-    * */
-  def test(testData: T): Seq[(I, Double, Double, Double, Double)] = {
-    logger.info("Generating predictions for test set")
-    //Calculate the posterior predictive distribution for the test points.
-    val predictionWithError = this.predictionWithErrorBars(dataAsIndexSeq(testData), 1)
-    //Collate the test data with the predictions and error bars
-    dataAsSeq(testData).zip(predictionWithError).map(i => (i._1._1, i._1._2,
-      i._2._2, i._2._3, i._2._4))
-  }
 
   /**
     * Cache the training kernel and noise matrices

dynaml-core/src/main/scala-2.11/io/github/mandar2812/dynaml/models/stp/AbstractSTPRegressionModel.scala

@@ -1,10 +1,9 @@
 package io.github.mandar2812.dynaml.models.stp
 
 import breeze.linalg.{DenseMatrix, DenseVector, cholesky}
-import breeze.numerics._
 import io.github.mandar2812.dynaml.kernels.LocalScalarKernel
-import io.github.mandar2812.dynaml.models.SecondOrderProcess
-import io.github.mandar2812.dynaml.optimization.{GloballyOptWithGrad, GloballyOptimizable}
+import io.github.mandar2812.dynaml.models.{ContinuousProcess, SecondOrderProcess}
+import io.github.mandar2812.dynaml.optimization.GloballyOptimizable
 import io.github.mandar2812.dynaml.probability.MultStudentsTRV
 import io.github.mandar2812.dynaml.probability.distributions.MultivariateStudentsT
 import org.apache.log4j.Logger
@@ -16,7 +15,8 @@ abstract class AbstractSTPRegressionModel[T, I](
   mu: Double, cov: LocalScalarKernel[I],
   n: LocalScalarKernel[I],
   data: T, num: Int)
-  extends SecondOrderProcess[T, I, Double, Double, DenseMatrix[Double], MultStudentsTRV]
+  extends ContinuousProcess[T, I, Double, MultStudentsTRV]
+  with SecondOrderProcess[T, I, Double, Double, DenseMatrix[Double], MultStudentsTRV]
   with GloballyOptimizable {
 
 
@@ -166,19 +166,6 @@ abstract class AbstractSTPRegressionModel[T, I](
     AbstractSTPRegressionModel.logLikelihood(current_state("degrees_of_freedom"),trainingLabels, kernelTraining)
   }
 
-  /**
-    * Returns a prediction with error bars for a test set of indexes and labels.
-    * (Index, Actual Value, Prediction, Lower Bar, Higher Bar)
-    * */
-  def test(testData: T): Seq[(I, Double, Double, Double, Double)] = {
-    logger.info("Generating predictions for test set")
-    //Calculate the posterior predictive distribution for the test points.
-    val predictionWithError = this.predictionWithErrorBars(dataAsIndexSeq(testData), 1)
-    //Collate the test data with the predictions and error bars
-    dataAsSeq(testData).zip(predictionWithError).map(i => (i._1._1, i._1._2,
-      i._2._2, i._2._3, i._2._4))
-  }
-
   /**
     * Cache the training kernel and noise matrices
     * for fast access in future predictions.