-------------------------------------------------
lines 15-284 of file: example/csv/sim_fit_pred.py
-------------------------------------------------

{xrst_begin csv.sim_fit_pred}
{xrst_spell
  const
  cv
  dage
  dtime
  iter
  meas
  mtexcess
  num
  pos
  sincidence
  std
}

Csv Example Simulating, Fitting and Predicting
##############################################

Global Variables
****************
Global variables besides sim_file and fit_file.
Variables that do not appear in a heading are temporaries.

integrand_list
==============
Generate data and predict for the following integrands:
{xrst_code py}'''
integrand_list = [ 'Sincidence', 'remission', 'mtexcess', 'prevalence' ]
'''{xrst_code}

age_grid, time_grid
===================
Use this age-time grid for values the covariate grid and the
parent rage grid.
{xrst_code py}'''
age_grid   = [0.0, 20.0, 50.0, 80.0, 100.0]
time_grid  = [1980.0, 2000.0, 2020.0]
'''{xrst_code}

node_dict
=========
Keys are nodes and values are corresponding parent node:
{xrst_code py}'''
node_dict = {
   'n0' : ''   ,
   'n1' : 'n0' ,
   'n2' : 'n0' ,
}
'''{xrst_code}

no_effect_rate_truth
====================
The true values (values used during simulation) for iota, rho, and chi
are constant w.r.t age and time:
{xrst_code py}'''
no_effect_rate_truth = {
   'iota' : 0.02  ,
   'rho'  : 20.0  ,
   'chi'  : 0.001  ,
}
'''{xrst_code}

omega_truth
===========
{xrst_code py}'''
omega_truth      = 0.01
'''{xrst_code}

std_random_effects_truth
========================
This is the true standard deviation of the random effects
{xrst_code py}'''
std_random_effects_truth = 0.2
'''{xrst_code}

sim_file
********
Input CSV files that are placed in the simulate directory:
{xrst_code py}'''
sim_file = dict()
'''{xrst_code}

option_sim.csv
==============
{xrst_code py}'''
sim_file['option_sim.csv'] = \
'''name,value
float_precision,4
random_depend_sex,false
'''
for rate_name in no_effect_rate_truth :
   row = f'std_random_effects_{rate_name},{std_random_effects_truth}\n'
   sim_file['option_sim.csv'] += row
'''{xrst_code}

node.csv
========
{xrst_code py}'''
sim_file['node.csv'] = \
'node_name,parent_name\n'
for node_name in node_dict :
   parent_name = node_dict[node_name]
   sim_file['node.csv'] += f'{node_name},{parent_name}\n'
'''{xrst_code}

covariate.csv
=============
{xrst_code py}'''
sim_file['covariate.csv'] = 'node_name,sex,age,time,omega\n'
for node_name in [ 'n0', 'n1', 'n2' ] :
   for sex in [ 'female', 'male' ] :
      for age in age_grid :
         for time in time_grid :
            row   = f'{node_name},{sex},{age},{time},{omega_truth}\n'
            sim_file['covariate.csv'] += row
'''{xrst_code}

multiplier_sim.csv
==================
There are no covariate multipliers in this example.
{xrst_code py}'''
sim_file['multiplier_sim.csv'] = \
   'multiplier_id,rate_name,covariate_or_sex,multiplier_true\n'
'''{xrst_code}

simulate.csv
============
{xrst_code py}'''
header  = 'simulate_id,integrand_name,node_name,sex,age_lower,age_upper,'
header += 'time_lower,time_upper,meas_std_cv,meas_std_min\n'
meas_std_cv     = 0.01
simulate_id     = 0
sim_file['simulate.csv'] = header
for integrand_name in integrand_list :
   std_min = 0.0
   if integrand_name == 'prevalence' :
      std_min = 1e-6
   for node_name in node_dict :
      for sex in [ 'female', 'male' ] :
         for age in age_grid :
            for time in time_grid :
               row  = f'{simulate_id},{integrand_name},{node_name},{sex},'
               row += f'{age},{age},{time},{time},'
               row += f'{meas_std_cv},{std_min}\n'
               sim_file['simulate.csv'] += row
               simulate_id += 1
'''{xrst_code}

no_effect_rate.csv
==================
The rates are constant, w.r.t age and time, during the simulation.
{xrst_code py}'''
sim_file['no_effect_rate.csv'] = 'rate_name,age,time,rate_truth\n'
for rate_name in no_effect_rate_truth :
   rate_truth = no_effect_rate_truth[rate_name]
   sim_file['no_effect_rate.csv'] += f'{rate_name},0,0,{rate_truth}\n'
'''{xrst_code}

fit_file
********
Input CSV files that are placed in the fit directory:
{xrst_code py}'''
fit_file = dict()
'''{xrst_code}

Copies of Simulation Files
==========================
{xrst_code py}'''
fit_file['node.csv']      = sim_file['node.csv']
fit_file['covariate.csv'] = sim_file['covariate.csv']
'''{xrst_code}

option_fit.csv
==============
{xrst_code py}'''
fit_file['option_fit.csv']  =  \
'''name,value
refit_split,false
ode_step_size,5.0
quasi_fixed,false
max_num_iter_fixed,50
tolerance_fixed,1e-8
ode_method,iota_pos_rho_pos
'''
'''{xrst_code}

option_predict.csv
==================
A predict is run using the same directory as the corresponding fit.
All of its input files are also inputs for the fit except for
the option_predict.csv file.
{xrst_code py}'''
fit_file['option_predict.csv']  =  \
'''name,value
db2csv,true
plot,true
float_precision,5
'''
'''{xrst_code}

fit_goal.csv
============
An empty fit_goal.csv corresponds to fitting all nodes at or below the
:ref:`root node<csv.fit@Input Files@option_fit.csv@root_node_name>` .
{xrst_code py}'''
fit_file['fit_goal.csv'] = \
'''node_name
'''
'''{xrst_code}

prior.csv
=========
{xrst_code py}'''
delta_prior_std        = 0.1
std_random_effects_fit = 10.0 * std_random_effects_truth
fit_file['prior.csv']  = \
   'name,density,mean,std,eta,lower,upper\n' + \
   f'delta_prior,log_gaussian,0.0,{delta_prior_std},1e-10,,\n' + \
   f'random_prior,gaussian,0.0,{std_random_effects_fit},,,,\n'
for rate_name in no_effect_rate_truth :
   rate_truth = no_effect_rate_truth[rate_name]
   lower      = rate_truth / 100.0
   upper      = rate_truth * 100.0
   fit_file['prior.csv'] += \
      f'prior_{rate_name},uniform,{rate_truth},,,{lower},{upper}\n'
'''{xrst_code}

parent_rate.csv
===============
The rates are constant during simulation, but not during fitting.
{xrst_code py}'''
fit_file['parent_rate.csv'] = \
   'rate_name,age,time,value_prior,dage_prior,dtime_prior,const_value\n'
for age in age_grid :
   for time in time_grid :
      for rate_name in no_effect_rate_truth :
         row  = f'{rate_name},{age},{time},prior_{rate_name},'
         row += 'delta_prior,delta_prior,\n'
         fit_file['parent_rate.csv'] += row
'''{xrst_code}

child_rate.csv
==============
{xrst_code py}'''
fit_file['child_rate.csv'] = 'rate_name,value_prior\n'
for rate_name in no_effect_rate_truth :
   fit_file['child_rate.csv'] += f'{rate_name},random_prior\n'
'''{xrst_code}

mulcov.csv
==========
{xrst_code py}'''
fit_file['mulcov.csv'] = 'covariate,type,effected,value_prior,const_value\n'
'''{xrst_code}

predict_integrand.csv
=====================
{xrst_code py}'''
fit_file['predict_integrand.csv'] = 'integrand_name\n'
for integrand_name in integrand_list :
   fit_file['predict_integrand.csv'] += f'{integrand_name}\n'
'''{xrst_code}

Rest of Source Code
*******************
{xrst_literal
   BEGIN PYTHON
   END PYTHON
}
{xrst_end csv.sim_fit_pred}