PACF plot (#62)

mj-xmr · Jan 26, 2022 · ca19c7a · ca19c7a
1 parent cb8a848
commit ca19c7a
Show file tree

Hide file tree

Showing 3 changed files with 25 additions and 6 deletions.
diff --git a/src/lib-base/src/MatplotACF.cpp b/src/lib-base/src/MatplotACF.cpp
@@ -36,7 +36,8 @@ void MatplotACF::Plot(const EnjoLib::VecD & dat, int lags, const EnjoLib::Str &
         AlgoSTDIVec<double>().Reverse(&reversed);
         //ofs << dat.PrintPython("dat") << Nl;
         ofs << reversed.PrintPython("dat") << Nl;
-        ofs << "plotData(dat, " << lags << ", '" << title << "')" << Nl;
+        ofs <<  "plotACF(dat, " << lags << ", '" << title << "')" << Nl;
+        ofs << "plotPACF(dat, " << lags << ", '" << title << "')" << Nl;
     }
 
     ToolsMixed().SystemCallWarn("python3 " + scriptOut, "MatplotACF::Plot()");

diff --git a/src/lib-base/src/OptimizerBase.cpp b/src/lib-base/src/OptimizerBase.cpp
@@ -123,7 +123,7 @@ void OptimizerBase::operator()()
             Consume(dataNewT.at(i));
             if (gcfgMan.cfgOpti->OPTI_RANDOM_EARLY_STOP && IsEarlyStop())
             {
-                LOGL << "Early stop. The recent changes were less than " << gcfgMan.cfgOpti->OPTI_RANDOM_MIN_DIFF_PROMILE << " ‰ after " << i << " iterations.\n";
+                LOGL << "Early stop. The recent variance changes were less than " << gcfgMan.cfgOpti->OPTI_RANDOM_MIN_DIFF_PROMILE << " ‰ after " << i << " iterations.\n";
                 break;
             }
         }
@@ -251,7 +251,7 @@ void OptimizerBase::PlotVariance() const
     const float scaleY = 0.5;
     const Formatters fmt;
     ELO
-    GnuplotPlotTerminal1d(m_goalsMod.Diffs().Smooth(GetSmoothing()), "Variance changes:", scaleX, scaleY);
+    GnuplotPlotTerminal1d(m_goalsMod.Diffs().Smooth(GetSmoothing()), "Variance changes (low values mean that no better solution can be found):", scaleX, scaleY);
     LOG << "Best  │  " << fmt.VecLabel() << "  │\n";
     LOG << fmt.FormatVar(m_goalMax) << " " << fmt.FormatVec(m_goals) << Endl;
     const StatsUtil::Distrib & distrib = StatsUtil().GetDistrib(m_goals);

diff --git a/src/tsqsim-lib/static/scripts/plot_acf_pacf.py b/src/tsqsim-lib/static/scripts/plot_acf_pacf.py
@@ -15,11 +15,28 @@
 #from pandas.plotting import register_matplotlib_converters
 #from statsmodels.graphics.tsaplots import plot_acf, plot_pacf
 #register_matplotlib_converters()
+from statsmodels.tsa.stattools import acf, pacf
 
 
-def plotData(data, lags):
+def plotACF(data, lags, title):
     obj = pd.plotting.autocorrelation_plot(data)
     obj.set_xlim([0, lags])
+    obj.set_title("Autocorr. (" + str(lags) + ") " + title)
+    plt.show()
+
+def plotPACF(data, lags, title):
+    pacf_vals = pacf(data)
+    #fig = plt.figure()
+    y = pacf_vals#[:lags]
+    length = len(y)
+    if lags > length:
+        lags = length
+    x = list(range(lags))
+    print(len(x))
+    print(length)
+    plt.bar(x, y)
+    plt.grid()
+    plt.title("Partial Autocorr. (" + str(lags) + ") " + title)
     plt.show()
 
 def demo():
@@ -41,12 +58,13 @@ def demo():
     #acf_plot = plot_acf(df_ice_cream.production, lags=100)
 
     a = [1, 2, 3, 4, 5, 6]
-    plotData(a, 4)
+    plotACF(a, 4, "demo")
+    plotPACF(a, 4, "demo")
     #obj = pd.plotting.autocorrelation_plot(df_ice_cream.production)
     #plt.show()
 
 
     #acf_plot = plot_acf(a, lags=5)
     #acf_plot.show()
 
-#demo()
+demo()