added diagnostic plots

darrenjw · Jun 4, 2017 · 4ddc9e0 · 4ddc9e0
1 parent 6935976
commit 4ddc9e0
Show file tree

Hide file tree

Showing 8 changed files with 158 additions and 2 deletions.
diff --git a/docs/api/index.js b/docs/api/index.js
diff --git a/docs/api/scalaglm/Glm.html b/docs/api/scalaglm/Glm.html
@@ -716,6 +716,22 @@ <h3>Value Members</h3>
 
       <p class="shortcomment cmt">p-values for the regression coefficients
 </p>
+    </li><li name="scalaglm.Glm#plots" visbl="pub" class="indented0 " data-isabs="false" fullComment="no" group="Ungrouped">
+      <a id="plots:breeze.plot.Figure"></a><a id="plots:Figure"></a>
+      <span class="permalink">
+      <a href="../scalaglm/Glm.html#plots:breeze.plot.Figure" title="Permalink">
+        <i class="material-icons"></i>
+      </a>
+    </span>
+      <span class="modifier_kind">
+        <span class="modifier"></span>
+        <span class="kind">def</span>
+      </span>
+      <span class="symbol">
+        <span class="name">plots</span><span class="result">: <span class="extype" name="breeze.plot.Figure">Figure</span></span>
+      </span>
+
+
     </li><li name="scalaglm.Glm#pp" visbl="pub" class="indented0 " data-isabs="false" fullComment="no" group="Ungrouped">
       <a id="pp:Int"></a>
       <span class="permalink">

diff --git a/docs/api/scalaglm/Lm.html b/docs/api/scalaglm/Lm.html
@@ -752,6 +752,22 @@ <h3>Value Members</h3>
 
       <p class="shortcomment cmt">The p-value associated with the f-statistic
 </p>
+    </li><li name="scalaglm.Lm#plots" visbl="pub" class="indented0 " data-isabs="false" fullComment="no" group="Ungrouped">
+      <a id="plots:breeze.plot.Figure"></a><a id="plots:Figure"></a>
+      <span class="permalink">
+      <a href="../scalaglm/Lm.html#plots:breeze.plot.Figure" title="Permalink">
+        <i class="material-icons"></i>
+      </a>
+    </span>
+      <span class="modifier_kind">
+        <span class="modifier"></span>
+        <span class="kind">def</span>
+      </span>
+      <span class="symbol">
+        <span class="name">plots</span><span class="result">: <span class="extype" name="breeze.plot.Figure">Figure</span></span>
+      </span>
+
+
     </li><li name="scalaglm.Lm#pp" visbl="pub" class="indented0 " data-isabs="false" fullComment="no" group="Ungrouped">
       <a id="pp:Int"></a>
       <span class="permalink">

diff --git a/docs/api/scalaglm/Pca.html b/docs/api/scalaglm/Pca.html
@@ -637,6 +637,23 @@ <h3>Value Members</h3>
       </span>
 
       <p class="shortcomment cmt">Number of variables
+</p>
+    </li><li name="scalaglm.Pca#plots" visbl="pub" class="indented0 " data-isabs="false" fullComment="no" group="Ungrouped">
+      <a id="plots:breeze.plot.Figure"></a><a id="plots:Figure"></a>
+      <span class="permalink">
+      <a href="../scalaglm/Pca.html#plots:breeze.plot.Figure" title="Permalink">
+        <i class="material-icons"></i>
+      </a>
+    </span>
+      <span class="modifier_kind">
+        <span class="modifier"></span>
+        <span class="kind">def</span>
+      </span>
+      <span class="symbol">
+        <span class="name">plots</span><span class="result">: <span class="extype" name="breeze.plot.Figure">Figure</span></span>
+      </span>
+
+      <p class="shortcomment cmt">Diagnostic plots for the PCA
 </p>
     </li><li name="scalaglm.Pca#propvar" visbl="pub" class="indented0 " data-isabs="false" fullComment="no" group="Ungrouped">
       <a id="propvar:breeze.linalg.DenseVector[Double]"></a><a id="propvar:DenseVector[Double]"></a>

diff --git a/docs/api/scalaglm/Utils$.html b/docs/api/scalaglm/Utils$.html
@@ -636,6 +636,42 @@ <h3>Value Members</h3>
       </span>
 
       <div class="fullcomment"><dl class="attributes block"> <dt>Definition Classes</dt><dd>AnyRef</dd></dl></div>
+    </li><li name="scalaglm.Utils#pairs" visbl="pub" class="indented0 " data-isabs="false" fullComment="yes" group="Ungrouped">
+      <a id="pairs(mat:breeze.linalg.DenseMatrix[Double]):breeze.plot.Figure"></a><a id="pairs(DenseMatrix[Double]):Figure"></a>
+      <span class="permalink">
+      <a href="../scalaglm/Utils$.html#pairs(mat:breeze.linalg.DenseMatrix[Double]):breeze.plot.Figure" title="Permalink">
+        <i class="material-icons"></i>
+      </a>
+    </span>
+      <span class="modifier_kind">
+        <span class="modifier"></span>
+        <span class="kind">def</span>
+      </span>
+      <span class="symbol">
+        <span class="name">pairs</span><span class="params">(<span name="mat">mat: <span class="extype" name="breeze.linalg.DenseMatrix">DenseMatrix</span>[<span class="extype" name="scala.Double">Double</span>]</span>)</span><span class="result">: <span class="extype" name="breeze.plot.Figure">Figure</span></span>
+      </span>
+
+      <p class="shortcomment cmt">pairs plot (for unlabeled variables)
+</p><div class="fullcomment"><div class="comment cmt"><p>pairs plot (for unlabeled variables)
+</p></div><dl class="paramcmts block"><dt class="param">mat</dt><dd class="cmt"><p>a matrix with variables in columns and observations in rows</p></dd><dt>returns</dt><dd class="cmt"><p>The breeze-viz Figure object</p></dd></dl></div>
+    </li><li name="scalaglm.Utils#pairs" visbl="pub" class="indented0 " data-isabs="false" fullComment="yes" group="Ungrouped">
+      <a id="pairs(mat:breeze.linalg.DenseMatrix[Double],names:Seq[String]):breeze.plot.Figure"></a><a id="pairs(DenseMatrix[Double],Seq[String]):Figure"></a>
+      <span class="permalink">
+      <a href="../scalaglm/Utils$.html#pairs(mat:breeze.linalg.DenseMatrix[Double],names:Seq[String]):breeze.plot.Figure" title="Permalink">
+        <i class="material-icons"></i>
+      </a>
+    </span>
+      <span class="modifier_kind">
+        <span class="modifier"></span>
+        <span class="kind">def</span>
+      </span>
+      <span class="symbol">
+        <span class="name">pairs</span><span class="params">(<span name="mat">mat: <span class="extype" name="breeze.linalg.DenseMatrix">DenseMatrix</span>[<span class="extype" name="scala.Double">Double</span>]</span>, <span name="names">names: <span class="extype" name="scala.Seq">Seq</span>[<span class="extype" name="scala.Predef.String">String</span>]</span>)</span><span class="result">: <span class="extype" name="breeze.plot.Figure">Figure</span></span>
+      </span>
+
+      <p class="shortcomment cmt">pairs plot
+</p><div class="fullcomment"><div class="comment cmt"><p>pairs plot
+</p></div><dl class="paramcmts block"><dt class="param">mat</dt><dd class="cmt"><p>a matrix with variables in columns and observations in rows</p></dd><dt class="param">names</dt><dd class="cmt"><p>a sequence of variable names</p></dd><dt>returns</dt><dd class="cmt"><p>The breeze-viz Figure object</p></dd></dl></div>
     </li><li name="scala.AnyRef#synchronized" visbl="pub" class="indented0 " data-isabs="false" fullComment="yes" group="Ungrouped">
       <a id="synchronized[T0](x$1:=&gt;T0):T0"></a><a id="synchronized[T0](⇒T0):T0"></a>
       <span class="permalink">

diff --git a/src/main/scala/scalaglm/Glm.scala b/src/main/scala/scalaglm/Glm.scala
@@ -172,6 +172,19 @@ case class Glm(y: DenseVector[Double],
 
   lazy val fitted = predict(response = true).fitted
 
+  import breeze.plot._
+  def plots: Figure = {
+    val fig = Figure("Linear regression diagnostics")
+    val p0 = fig.subplot(1,1,0)
+    p0 += plot(fitted,y,'+')
+    p0 += plot(fitted,fitted)
+    p0.title = "Observations against fitted values"
+    p0.xlabel = "Fitted value"
+    p0.ylabel = "Observation"
+    fig
+  }
+
+
 } // case class Glm
 
 object Glm {

diff --git a/src/main/scala/scalaglm/Lm.scala b/src/main/scala/scalaglm/Lm.scala
@@ -212,7 +212,22 @@ case class Lm(y: DenseVector[Double],
   import breeze.plot._
   def plots: Figure = {
     val fig = Figure("Linear regression diagnostics")
-
+    val p0 = fig.subplot(2,2,0)
+    p0 += plot(fitted,y,'+')
+    p0 += plot(fitted,fitted)
+    p0.title = "Observations against fitted values"
+    p0.xlabel = "Fitted value"
+    p0.ylabel = "Observation"
+    val p1 = fig.subplot(2,2,1)
+    p1 += plot(fitted,residuals,'+')
+    p1.title = "Residuals against fitted values"
+    p1.xlabel = "Fitted value"
+    p1.ylabel = "Residual"
+    val p2 = fig.subplot(2,2,2)
+    p2 += breeze.plot.hist(residuals)
+    p2.title = "Histogram of residuals"
+    p2.xlabel = "Residual"
+    // TODO: Add a Q-Q plot..
     fig
   }
 

diff --git a/src/main/scala/scalaglm/Utils.scala b/src/main/scala/scalaglm/Utils.scala
@@ -119,6 +119,49 @@ object Utils {
   }
 
 
+  import breeze.plot._
+  /** 
+    * pairs plot
+    * 
+    * @param mat a matrix with variables in columns and observations in rows
+    * @param names a sequence of variable names
+    * 
+    * @return The breeze-viz Figure object
+    */
+  def pairs(mat: DenseMatrix[Double], names: Seq[String]): Figure = {
+    require(mat.cols == names.length)
+    val fig = Figure("Scatterplot matrix")
+    val p = mat.cols
+      (0 until p).foreach{i =>
+        (0 until p).foreach {j =>
+          val pij = fig.subplot(p,p,i*p+j)
+          if (i == j) {
+            pij += hist(mat(::,i))
+            pij.title = names(i)
+            pij.xlabel = i.toString
+          } else {
+            pij += plot(mat(::,j),mat(::,i),'.')
+            pij.xlabel = j.toString
+            pij.ylabel = i.toString
+          }
+        }
+      }
+    fig
+  }
+
+  /** 
+    * pairs plot (for unlabeled variables)
+    * 
+    * @param mat a matrix with variables in columns and observations in rows
+    * 
+    * @return The breeze-viz Figure object
+    */
+  def pairs(mat: DenseMatrix[Double]): Figure = {
+    val names = (1 to mat.cols) map ("V%02d".format(_))
+    pairs(mat, names)
+  }
+
+
   /** 
     * Example of a main runner method - not for general use - will probably get removed
     * in due course