EPNM paper review (#394)

Author: twallema

.github/workflows/deploy.yml

@@ -14,14 +14,14 @@ jobs:
     - name: Set up Python
       uses: actions/setup-python@v1
       with:
-        python-version: '3.7'
+        python-version: '3.12'
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
         pip install -e .[develop]
     - name: Build documentation
       run: |
-        python setup.py build_sphinx
+        sphinx-build docs build/sphinx/
     - name: Deploy documentation
       uses: peaceiris/actions-gh-pages@v3
       with:

.github/workflows/tests.yml

@@ -14,7 +14,7 @@ jobs:
     - name: Set up Python
       uses: actions/setup-python@v1
       with:
-        python-version: '3.7'
+        python-version: '3.12'
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip

.gitignore

@@ -170,3 +170,6 @@ data/covid19_DTM/interim/QALY_model/postponement_non_covid_care/UZG/MZG_2016_202
 data/covid19_DTM/interim/QALY_model/postponement_non_covid_care/UZG/2020_2021_normalized.csv
 # Do not upload COVID19-Model.code-workspace for VS Studio
 COVID19-Model.code-workspace
+
+# Macbook
+.DS_Store

notebooks/manuscripts/EPNM/plot-fit-sectors-EPNM.py

@@ -1,15 +1,11 @@
 # General packages
-import os
 import numpy as np
 import pandas as pd
 import matplotlib.pyplot as plt
 # EPNM functions
 from EPNM.models.utils import initialize_model
-from EPNM.data.parameters import get_model_parameters
 from EPNM.data.utils import get_sector_labels, get_sector_names, aggregate_simulation, get_sectoral_conversion_matrix
-from EPNM.models.TDPF import household_demand_shock, compute_income_expectations
-from EPNM.models.draw_functions import draw_function as draw_function
-from EPNM.data.calibration_data import get_NAI_value_added, get_revenue_survey, get_employment_survey, get_synthetic_GDP, get_B2B_demand
+from EPNM.data.calibration_data import get_revenue_survey, get_employment_survey, get_synthetic_GDP, get_B2B_demand
 # Nicer colors
 colors = {"orange" : "#E69F00", "light_blue" : "#56B4E9",
           "green" : "#009E73", "yellow" : "#F0E442",
@@ -36,6 +32,7 @@
 # Initialize model
 params, model = initialize_model(shocks='pichler', prodfunc='half_critical')
 
+# Pichler parameters
 model.parameters.update({'gamma_F': 15, 'gamma_H': 30, 'tau': 10, 'rho': 1-(1-0.60)/90})
 
 # Aggregation functions
@@ -73,9 +70,6 @@ def aggregate_NACE21(simulation_in):
                                     date=simulation_in.coords['date']))
     return simulation_out
 
-# Draw function
-from EPNM.models.draw_functions import draw_function
-
 # Simulate
 out = model.sim([start_sim, end_sim], tau=1)
 
@@ -132,7 +126,7 @@ def aggregate_NACE21(simulation_in):
             dist_temp.append(B2B_demand[j]/sum(B2B_demand)*(abs(x)-abs(y.values)) )
             # Sector label
             if sector in ['I', 'R', 'S']:
-                ax[0,i].annotate(sector,  xy=(x - 2, y + 2), fontsize=7)
+                ax[0,i].annotate(sector,  xy=(x - 2, y + 2), fontsize=6)
             # circle around transport
             if sector == 'H':
                 from matplotlib import pyplot as plt, patches
@@ -161,8 +155,8 @@ def aggregate_NACE21(simulation_in):
 datasets = [data_GDP, data_revenue, data_employment]
 states = ['x', 'x', 'l']
 sizes = [params['x_0'], params['x_0'], params['l_0']]
-print_label = [['G45',],['G45','N79', 'R93'],['G45','N79', 'R93']]
-offset = [[-4,5],[-4,5],[-4,5]]
+print_label = [['G45', 'H49', 'N77'],['G45','N79', 'R93','I55-56', 'H49'],['G45','N79', 'R93','I55-56', 'H49']]
+offset = [[-5,5],[-5,5],[-5,5]]
 ylabels = ['Synthetic GDP\nprediction (%)', 'Revenue\nprediction (%)', 'Employment\nprediction (%)']
 
 #########
@@ -205,7 +199,14 @@ def aggregate_NACE21(simulation_in):
                 cumsize.append(sizes[k][get_sector_labels('NACE64').index(sector)]/sum(sizes[k]))
                 # Sector label
                 if sector in print_label[k]:
-                    ax[k+1,i].annotate(sector,  xy=(x + offset[k][0], y + offset[k][1]), fontsize=7)
+                    if ((k == 1) & (i == 0) & (sector == 'I55-56')):
+                        pass
+                    elif ((k == 2) & (i == 0) & (sector == 'I55-56')):
+                        pass
+                    elif ((k == 2) & (i == 2) & (sector == 'I55-56')):
+                        pass
+                    else:
+                        ax[k+1,i].annotate(sector,  xy=(x + offset[k][0], y + offset[k][1]), fontsize=6)
                 # Circle around transport
                 if ((sector == 'H49')):
                     from matplotlib import pyplot as plt, patches

notebooks/manuscripts/EPNM/plot-fit-sensivity.py

@@ -29,30 +29,30 @@
 ## Define draw functions ##
 ###########################
 
-def draw_function_b(param_dict, samples_dict):
-    param_dict['b_s'] = np.random.uniform(low=0.5, high=1)
-    return param_dict
+def draw_function_b(parameters):
+    parameters['b_s'] = np.random.uniform(low=0.5, high=1)
+    return parameters
 
-def draw_function_L(param_dict, samples_dict):
-    param_dict['L'] = np.random.uniform(low=0.5, high=1)
-    return param_dict
+def draw_function_L(parameters):
+    parameters['L'] = np.random.uniform(low=0.5, high=1)
+    return parameters
 
-def draw_function_delta_S(param_dict, samples_dict):
-    param_dict['delta_S'] = np.random.uniform(low=0.5, high=1)
-    return param_dict
+def draw_function_delta_S(parameters):
+    parameters['delta_S'] = np.random.uniform(low=0.5, high=1)
+    return parameters
 
-def draw_function_rho(param_dict, samples_dict):
-    param_dict['rho'] = np.random.uniform(low=0.99, high=0.999) # 0.1 to 1 quarter
-    return param_dict
+def draw_function_rho(parameters):
+    parameters['rho'] = np.random.uniform(low=0.99, high=0.999) # 0.1 to 1 quarter
+    return parameters
 
-def draw_function_l_s(param_dict, samples_dict):
+def draw_function_l_s(parameters):
     r = np.random.uniform(low=0.70, high=1.30)
-    param_dict['l_s_1'] =  r*param_dict['l_s_1']
-    param_dict['l_s_2'] =  r*param_dict['l_s_2']
-    return param_dict
+    parameters['l_s_1'] =  r*parameters['l_s_1']
+    parameters['l_s_2'] =  r*parameters['l_s_2']
+    return parameters
 
 draw_functions = [draw_function_l_s, draw_function_b, draw_function_rho, draw_function_L, draw_function_delta_S]
-parameter_names = ['$\\epsilon_{i,t}^S$', '$b_t$', '$\\rho$', '$L$', '$\Delta S$']
+parameter_names = ['$\\kappa_{i,t}^S$', '$b_t$', '$\\rho$', '$L$', '$\\Delta S$']
 
 ######################
 ## Initialize model ##
@@ -69,7 +69,7 @@ def draw_function_l_s(param_dict, samples_dict):
 
 for i,draw_function in enumerate(draw_functions):
     # Simulate model
-    out = model.sim([start_sim, end_sim], method='RK45', rtol=1e-4, N=5*18, processes=18, samples={}, draw_function=draw_function)
+    out = model.sim([start_sim, end_sim], method='RK45', rtol=1e-4, N=5*16, processes=16, draw_function=draw_function)
     simtime = out['date'].values
     # Skip row 0, column 1
 

notebooks/manuscripts/EPNM/plot_sensitivity-grid.ipynb

@@ -21,6 +21,7 @@
         "develop":  ["pytest",
                      "sphinx",
                      "numpydoc",
+                     "setuptools",
                      "sphinx_rtd_theme",
                      "myst_parser[sphinx]"],
     }

notebooks/manuscripts/EPNM/timeline.ipynb

@@ -1,32 +1,32 @@
 import numpy as np
 
-def draw_function(param_dict, samples_dict):
+def draw_function(parameters):
     """
     Samples calibrated magnitudes of household demand/exogeneous demand shocks during the summer of 2021
     Performs global sensitivity analysis on other parameters
     """
 
     # Option 1
     # shocks in/out
-    param_dict['l1'] = np.random.normal(loc=14, scale=0.2*14)
-    param_dict['l2'] = np.random.normal(loc=7*8, scale=0.2*7*8)
+    parameters['l1'] = np.random.normal(loc=14, scale=0.2*14)
+    parameters['l2'] = np.random.normal(loc=7*8, scale=0.2*7*8)
     # hiring and Firing speed
-    param_dict['gamma_F'] = np.random.normal(loc=28, scale=0.2*28)
-    param_dict['gamma_H'] = 2*param_dict['gamma_F']
+    parameters['gamma_F'] = np.random.normal(loc=28, scale=0.2*28)
+    parameters['gamma_H'] = 2*parameters['gamma_F']
     # household savings and prospects
-    param_dict['delta_S'] = np.random.uniform(low=0.5, high=1)
-    param_dict['L'] = 1
-    param_dict['b_s'] = 0.7
+    parameters['delta_S'] = np.random.uniform(low=0.5, high=1)
+    parameters['L'] = 1
+    parameters['b_s'] = 0.7
     # restitution 
-    param_dict['rho'] = np.random.uniform(low=1-(1-0.50)/90, high=1-(1-1)/90)
+    parameters['rho'] = np.random.uniform(low=1-(1-0.50)/90, high=1-(1-1)/90)
     # restock rate
-    param_dict['tau'] =  np.random.normal(loc=7, scale=0.2*7)
+    parameters['tau'] =  np.random.normal(loc=7, scale=0.2*7)
     # shocks lockdowns
-    param_dict['l_s_1'] =  np.random.normal(loc=1, scale=0.075)*param_dict['l_s_1']
-    param_dict['l_s_2'] =  np.random.normal(loc=1, scale=0.075)*param_dict['l_s_2']
-    param_dict['c_s'] =  np.random.normal(loc=1, scale=0.075)*param_dict['c_s']
-    param_dict['c_s'] = np.where(param_dict['c_s'] > 1, 1, param_dict['c_s'])
+    parameters['l_s_1'] =  np.random.normal(loc=1, scale=0.075)*parameters['l_s_1']
+    parameters['l_s_2'] =  np.random.normal(loc=1, scale=0.075)*parameters['l_s_2']
+    parameters['c_s'] =  np.random.normal(loc=1, scale=0.075)*parameters['c_s']
+    parameters['c_s'] = np.where(parameters['c_s'] > 1, 1, parameters['c_s'])
     # shocks summer 2020 
-    param_dict['ratio_c_s'] =  np.random.uniform(low=0.25, high=0.75)
+    parameters['ratio_c_s'] =  np.random.uniform(low=0.25, high=0.75)
 
-    return param_dict
+    return parameters