Merge pull request #26 from Plant-Food-Research-Open/Jfruit2_SA

Add sensitivity analysis notebook for Jfruit2

Author: JBris

workshop/2_jfruit2/1_jfruit2_sensitivity_analysis.ipynb

@@ -50,7 +50,7 @@
   },
   {
    "cell_type": "code",
-   "execution_count": 1,
+   "execution_count": null,
    "id": "afa8f0e4-8216-44fd-8770-fbeec0b388cf",
    "metadata": {},
    "outputs": [],
@@ -600,15 +600,85 @@
    "id": "14de1930-423b-4c1b-a1cb-d5b0c8d6f6d0",
    "metadata": {},
    "source": [
-    "# Results\n",
+    "# Results"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "53eab109",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "print(\"JFruit2 sensitivity analysis\")\n",
+    "print(\"=\" * 60)\n",
+    "\n",
+    "# Get results from the calibrator\n",
+    "sp = calibrator.implementation.sp\n",
+    "Y = sp._results\n",
+    "Si = sobol.analyze(sp, Y, calc_second_order=True)\n",
+    "\n",
+    "output_labels = calibrator.specification.output_labels[0]  \n",
+    "\n",
+    "s1_values = Si['S1']\n",
+    "st_values = Si['ST']\n",
+    "parameter_names = sp['names']\n",
+    "\n",
+    "# results summary\n",
+    "sensitivity_df = pd.DataFrame({\n",
+    "    'Parameter': parameter_names,\n",
+    "    'S1 (direct effect)': s1_values,\n",
+    "    'ST (total effect)': st_values,\n",
+    "    'Interactions': st_values - s1_values\n",
+    "}).sort_values('ST (total effect)', ascending=False)\n",
+    "\n",
+    "print(sensitivity_df.to_string(index=False, float_format='%.3f'))\n",
+    "\n",
+    "print(\"=\" * 60)\n",
+    "for _, row in sensitivity_df.iterrows():\n",
+    "    param = row['Parameter']\n",
+    "    s1 = row['S1 (direct effect)']\n",
+    "    st = row['ST (total effect)']\n",
+    "    \n",
+    "    print(f\"\\n• {param}: {st:.1%} total influence on {output_labels}\")\n",
+    "    \n",
+    "    if 'Growth.lp' in param:\n",
+    "        print(f\"  - Controls how quickly the fruit grows\")\n",
+    "    elif 'Growth.lxAsy' in param:\n",
+    "        print(f\"  - Sets the maximum potential fruit size\")\n",
+    "    \n",
+    "    if st > 0.1:\n",
+    "        print(f\"  - 🔴 HIGH importance - prioritize for calibration\")\n",
+    "    elif st > 0.05:\n",
+    "        print(f\"  - 🟡 MEDIUM importance - secondary calibration target\")\n",
+    "    else:\n",
+    "        print(f\"  - 🟢 LOW importance - can use literature values\")\n",
+    "\n",
+    "\n",
+    "print(\"=\" * 60)\n",
+    "most_important = sensitivity_df.iloc[0]\n",
+    "max_interaction = sensitivity_df['Interactions'].max()\n",
+    "\n",
+    "print(f\"• Most important parameter: {most_important['Parameter']} ({most_important['ST (total effect)']:.1%} influence)\")\n",
+    "print(f\"• Parameter interactions: {'Strong' if max_interaction > 0.1 else 'Weak'}\")\n",
+    "\n",
+    "if max_interaction > 0.1:\n",
+    "    print(f\"  → Calibrate parameters together, not separately\")\n",
+    "else:\n",
+    "    print(f\"  → Parameters can be calibrated independently\")\n",
     "\n",
-    "According to the plots above..."
+    "total_explained = np.sum(st_values)\n",
+    "print(f\"• Model variance explained: {total_explained:.1%}\")\n",
+    "if total_explained > 0.8:\n",
+    "    print(f\"  → The parameters explain most model behaviours in the response variables\")\n",
+    "else:\n",
+    "    print(f\"  → The model outputs are not sensitive to the parameter set selected\")"
    ]
   }
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3 (ipykernel)",
+   "display_name": "Python 3",
    "language": "python",
    "name": "python3"
   },
@@ -622,7 +692,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.17"
+   "version": "3.10.5"
   }
  },
  "nbformat": 4,