CircleCI build openturns 18192

openturns/master/_downloads/03cf4bff12875027bc76f689658b754b/plot_estimate_gev_pirie.ipynb

@@ -166,7 +166,7 @@
       },
       "outputs": [],
       "source": [
-        "result_PLL = factory.buildMethodOfProfileLikelihoodMaximizationEstimator(sample)"
+        "result_PLL = factory.buildMethodOfXiProfileLikelihoodEstimator(sample)"
       ]
     },
     {
@@ -292,7 +292,7 @@
       },
       "outputs": [],
       "source": [
-        "constant = ot.SymbolicFunction([\"t\"], [\"1.0\"])\nbasis_lin = ot.Basis([constant, ot.SymbolicFunction([\"t\"], [\"t\"])])\nbasis_cst = ot.Basis([constant])\n# basis for mu, sigma, xi\nbasis_coll = [basis_lin, basis_cst, basis_cst]"
+        "constant = ot.SymbolicFunction([\"t\"], [\"1.0\"])\nbasis = ot.Basis([constant, ot.SymbolicFunction([\"t\"], [\"t\"])])\n# basis for mu, sigma, xi\nmuIndices = [0, 1]  # linear\nsigmaIndices = [0]  # stationary\nxiIndices = [0]  # stationary"
       ]
     },
     {
@@ -328,7 +328,7 @@
       },
       "outputs": [],
       "source": [
-        "initiPoint_list = list()\ninitiPoint_list.append(\"Gumbel\")\ninitiPoint_list.append(\"Static\")\nnormMethod_list = list()\nnormMethod_list.append(\"MinMax\")\nnormMethod_list.append(\"CenterReduce\")\nnormMethod_list.append(\"None\")\nprint(\"Linear mu(t) model: \")\nfor normMeth in normMethod_list:\n    for initPoint in initiPoint_list:\n        print(\"normMeth, initPoint = \", normMeth, initPoint)\n        # The ot.Function() is the identity function.\n        result = factory.buildTimeVarying(\n            sample, timeStamps, basis_coll, ot.Function(), initPoint, normMeth\n        )\n        beta = result.getOptimalParameter()\n        print(\"beta1, beta2, beta3, beta4 = \", beta)\n        print(\"Max log-likelihood =  \", result.getLogLikelihood())"
+        "print(\"Linear mu(t) model: \")\nfor normMeth in [\"MinMax\", \"CenterReduce\", \"None\"]:\n    for initPoint in [\"Gumbel\", \"Static\"]:\n        print(f\"normMeth = {normMeth}, initPoint = {initPoint}\")\n        # The ot.Function() is the identity function.\n        result = factory.buildTimeVarying(\n            sample, timeStamps, basis, muIndices, sigmaIndices, xiIndices,\n            ot.Function(), ot.Function(), ot.Function(), initPoint, normMeth\n        )\n        beta = result.getOptimalParameter()\n        print(f\"beta = {beta}\")\n        print(f\"Max log-likelihood = {result.getLogLikelihood()}\")"
       ]
     },
     {
@@ -346,7 +346,7 @@
       },
       "outputs": [],
       "source": [
-        "result_NonStatLL = factory.buildTimeVarying(sample, timeStamps, basis_coll)\nbeta = result_NonStatLL.getOptimalParameter()\nprint(\"beta1, beta2, beta3, beta_4 = \", beta)\nprint(f\"mu(t) = {beta[0]:.4f} + {beta[1]:.4f} * tau\")\nprint(f\"sigma = {beta[2]:.4f}\")\nprint(f\"xi = {beta[3]:.4f}\")"
+        "result_NonStatLL = factory.buildTimeVarying(sample, timeStamps, basis, muIndices, sigmaIndices, xiIndices)\nbeta = result_NonStatLL.getOptimalParameter()\nprint(f\"beta = {beta}\")\nprint(f\"mu(t) = {beta[0]:.4f} + {beta[1]:.4f} * tau\")\nprint(f\"sigma = {beta[2]:.4f}\")\nprint(f\"xi = {beta[3]:.4f}\")"
       ]
     },
     {
@@ -425,7 +425,7 @@
       "cell_type": "markdown",
       "metadata": {},
       "source": [
-        "In order to draw some diagnostic plots similar to those drawn in the stationary case, we refer to the\nfollowing result: if $Z_t$ is a non stationary GEV model parametrized by $(\\mu(t), \\sigma(t), \\xi(t))$,\nthen the standardized variables $\\hat{Z}_t$ defined by:\n\n\\begin{align}\\hat{Z}_t = \\dfrac{1}{\\xi(t)} \\log \\left[1+ \\xi(t)\\left( \\dfrac{Z_t-\\mu(t)}{\\sigma(t)} \\right)\\right]\\end{align}\n\nhave  the standard Gumbel distribution which is the GEV model with $(\\mu, \\sigma, \\xi) = (0, 1, 0)$.\n\nAs a result, we can validate the inference result thanks the 4 usual diagnostic plots:\n\n- the probability-probability pot,\n- the quantile-quantile pot,\n- the return level plot,\n- the data histogram and the desnity of the fitted model.\n\nusing the transformed data compared to the Gumbel model. We can see that the adequation seems similar to the graph\nof the stationary model.\n\n"
+        "In order to draw some diagnostic plots similar to those drawn in the stationary case, we refer to the\nfollowing result: if $Z_t$ is a non stationary GEV model parametrized by $(\\mu(t), \\sigma(t), \\xi(t))$,\nthen the standardized variables $\\hat{Z}_t$ defined by:\n\n\\begin{align}\\hat{Z}_t = \\dfrac{1}{\\xi(t)} \\log \\left[1+ \\xi(t)\\left( \\dfrac{Z_t-\\mu(t)}{\\sigma(t)} \\right)\\right]\\end{align}\n\nhave the standard Gumbel distribution which is the GEV model with $(\\mu, \\sigma, \\xi) = (0, 1, 0)$.\n\nAs a result, we can validate the inference result thanks the 4 usual diagnostic plots:\n\n- the probability-probability pot,\n- the quantile-quantile pot,\n- the return level plot,\n- the data histogram and the density of the fitted model.\n\nusing the transformed data compared to the Gumbel model. We can see that the adequation seems similar to the graph\nof the stationary model.\n\n"
       ]
     },
     {
@@ -454,7 +454,7 @@
       },
       "outputs": [],
       "source": [
-        "graph = ot.Graph(\n    r\"Maximum annual sea-levels at Port Pirie - Linear $\\mu(t)$\",\n    \"year\",\n    \"level (m)\",\n    True,\n    \"\",\n)\ngraph.setIntegerXTick(True)\n# data\ncloud = ot.Cloud(data[:, :2])\ncloud.setColor(\"red\")\ngraph.add(cloud)\n# mean function\nmeandata = [\n    result_NonStatLL.getDistribution(t).getMean()[0] for t in data[:, 0].asPoint()\n]\ncurve_meanPoints = ot.Curve(data[:, 0].asPoint(), meandata)\ngraph.add(curve_meanPoints)\n# quantile function\ngraphQuantile = result_NonStatLL.drawQuantileFunction(0.95)\ndrawQuant = graphQuantile.getDrawable(0)\ndrawQuant = graphQuantile.getDrawable(0)\ndrawQuant.setLineStyle(\"dashed\")\ngraph.add(drawQuant)\ngraph.setLegends([\"data\", \"mean function\", \"quantile 0.95  function\"])\ngraph.setLegendPosition(\"lower right\")\nview = otv.View(graph)"
+        "graph = ot.Graph(\n    r\"Maximum annual sea-levels at Port Pirie - Linear $\\mu(t)$\",\n    \"year\",\n    \"level (m)\",\n    True,\n    \"\",\n)\ngraph.setIntegerXTick(True)\n# data\ncloud = ot.Cloud(data[:, :2])\ncloud.setColor(\"red\")\ngraph.add(cloud)\n# mean function\nmeandata = [\n    result_NonStatLL.getDistribution(t).getMean()[0] for t in data[:, 0].asPoint()\n]\ncurve_meanPoints = ot.Curve(data[:, 0].asPoint(), meandata)\ngraph.add(curve_meanPoints)\n# quantile function\ngraphQuantile = result_NonStatLL.drawQuantileFunction(0.95)\ndrawQuant = graphQuantile.getDrawable(0)\ndrawQuant = graphQuantile.getDrawable(0)\ndrawQuant.setLineStyle(\"dashed\")\ngraph.add(drawQuant)\ngraph.setLegends([\"data\", \"mean function\", \"quantile 0.95 function\"])\ngraph.setLegendPosition(\"lower right\")\nview = otv.View(graph)"
       ]
     },
     {