import pandas as pd

import numpy as np

from pymc3 import Model, Normal, Gamma, summary, Uniform, traceplot, Slice, Metropolis, sample, find_MAP

import math

import scipy as sp

from scipy import optimize

#importing modules containing functions

import temp_functions_all as tfa

import mcmc_utils_all as mua

#importing plotting functions

import matplotlib.pyplot as plt

figure_count= 0

data_all= pd.read_csv('aegyptiDENVmodelTempData_2016-03-30.csv')

# Exclude the Focks & Barrera 2006 data because they're from a model

data_all = data_all.query('ref!= "Focks_Barrera_2006_Research&TrainingTropicalDis_Geneva_Paper"')

# PYMC parameters

sample_size= 5000

#############################################

################# GCR #######################

#############################################

#pEA, b,

data = data_all.query('trait_name== "GCR"')

#GET THE VALUES

Y = data['trait']

Y=Y.as_matrix()

Y= Y.flatten()

#Get the temperatures

T= data['T']

T=T.as_matrix()

T= T.flatten()

size= len(Y)

#TRANSFORM FROM DATAFRAME TO NP ARRAY FORMAT

# Specifing the parameters that control the MCMC (these will be used throughout the code).

basic_model_GCR = Model()

with basic_model_GCR:

#priors for unknown model parameters

c = Gamma('c',1,10)

Tm= Uniform('Tm',25,45)

T0= Uniform('T0',0,24)

tau= Gamma('tau',0.0001, 0.0001)

mu_temp= c*T*((T-T0)*(T0<T))*np.sqrt((Tm-T)*(Tm>T))

mu= 0*(mu_temp<0) + mu_temp*(mu_temp>0)

Y_obs = Normal('Y_obs',mu=mu, sd=tau, observed= Y)

from pymc3 import Metropolis, sample, find_MAP

from scipy import optimize

with basic_model_GCR:

# obtain starting values via MAP

start = find_MAP(fmin=optimize.fmin_powell)

# draw 5000 posterior samples

trace= sample(sample_size, step= Metropolis(), start=start)

#thin the samples by selecting every 5 samples

thin_factor=5

#summary(trace)

#traceplot(trace);

#PLOTTING THE HISTOGRAM

#figure_count= mua.create_2x2_histograms(trace, figure_count)

#Create the Brier Function

Temps= np.arange(0,50,0.1)

a_samps= mua.make_sims_temp_resp("briere",trace, Temps, thin_factor )

samps={}

samps['GCR']=a_samps

samps['T']=Temps

q= mua.temp_sim_quants(a_samps,Temps)

#PLOTTING THE DATA

#figure_count= mua.add_sim_lines(Temps, a_samps, q,Y , T, figure_count)

#CHECK COUNT:*

#############################################

################# EFD #######################

#############################################

#pEA, b,

data = data_all.query('trait_name== "EFD"')

#GET THE VALUES

Y = data['trait']

Y=Y.as_matrix()

Y= Y.flatten()

#Get the temperatures

T= data['T']

T=T.as_matrix()

T= T.flatten()

size= len(Y)

#TRANSFORM FROM DATAFRAME TO NP ARRAY FORMAT

# Specifing the parameters that control the MCMC (these will be used throughout the code).

basic_model_EFD = Model()

with basic_model_EFD:

#priors for unknown model parameters

c = Gamma('c',9,1000)

Tm= Uniform('Tm',25,45)

T0= Uniform('T0',0,24)

tau= Gamma('tau',10, 1)

mu_temp= c*T*((T-T0)*(T0<=T))*np.sqrt((Tm-T)*(Tm>=T))

mu= 0*(mu_temp<0) + mu_temp*(mu_temp>0)

Y_obs = Normal('Y_obs',mu=mu, sd=tau, observed= Y)

with basic_model_EFD:

# obtain starting values via MAP

start = find_MAP(fmin=optimize.fmin_powell)

# draw 5000 posterior samples

trace= sample(sample_size, step= Metropolis(), start=start)

#thin the samples by selecting every 5 samples

thin_factor=5

#summary(trace)

#traceplot(trace);

#PLOTTING THE HISTOGRAM

#figure_count= mua.create_2x2_histograms(trace, figure_count)

#Create the Brier Function

Temps= np.arange(0,50,0.1)

EFD_samps= mua.make_sims_temp_resp("briere",trace, Temps, thin_factor )

samps['EFD']=EFD_samps

q= mua.temp_sim_quants(EFD_samps,Temps)

#PLOTTING THE DATA

#figure_count= mua.add_sim_lines(Temps, EFD_samps, q,Y , T, figure_count)

q= mua.temp_sim_quants(EFD_samps,Temps)

#PLOTTING THE DATA

#figure_count= mua.add_sim_lines(Temps, EFD_samps, q,Y , T, figure_count)

#CHECK COUNT:*

#############################################

################# b #######################

#############################################

data = data_all.query('trait_name== "b"')

data= data.query('ref!="Lambrects_et_al_2011PNAS"')

#GET THE VALUES

Y = data['trait']

Y=Y.as_matrix()

Y= Y.flatten()

#Get the temperatures

T= data['T']

T=T.as_matrix()

T= T.flatten()

size= len(Y)

#TRANSFORM FROM DATAFRAME TO NP ARRAY FORMAT

basic_model_b= Model()

with basic_model_b:

#priors for unknown model parameters

c = Gamma('c',7,10000)

Tm= Uniform('Tm',25,45)

T0= Uniform('T0',0,24)

tau= Gamma('tau',10, 1)

mu_temp= c*T*((T-T0)*(T0<T))*np.sqrt((Tm-T)*(Tm>=T))

mu= 1*(mu_temp>=1) + mu_temp*(mu_temp>0)*(mu_temp<1) + 0*(mu_temp<=0)

Y_obs = Normal('Y_obs',mu=mu, sd=tau, observed= Y)

# Specifing the parameters that control the MCMC (these will be used throughout the code).

with basic_model_b:

# obtain starting values via MAP

start = find_MAP(fmin=optimize.fmin_powell)

# draw 5000 posterior samples

trace= sample(sample_size, step= Metropolis(), start=start)

#thin the samples by selecting every 5 samples

thin_factor=5

#PLOTTING THE HISTOGRAM

#figure_count= mua.create_2x2_histograms(trace, figure_count)

#Create the Brier Function

Temps= np.arange(0,50,0.1)

b_samps= mua.make_sims_temp_resp("briere_trunc",trace, Temps, thin_factor )

samps['b']=b_samps

#calculate the quants

q= mua.temp_sim_quants(b_samps,Temps)

#PLOTTING THE DATA

#figure_count= mua.add_sim_lines(Temps, b_samps, q,Y , T, figure_count)

#CHECK COUNT:*

#############################################

################# c #######################

#############################################

data = data_all.query('trait_name== "c"')

data= data.query('ref!="Lambrects_et_al_2011PNAS"')

data= data.query('ref!="Alto&Bettinardi_2013_AJTMH"')

#GET THE VALUES

Y = data['trait']

Y=Y.as_matrix()

Y= Y.flatten()

#Get the temperatures

T= data['T']

T=T.as_matrix()

T= T.flatten()

size= len(Y)

#TRANSFORM FROM DATAFRAME TO NP ARRAY FORMAT

basic_model_c= Model()

with basic_model_c:

#priors for unknown model parameters

c = Gamma('c',1,10)

Tm= Uniform('Tm',25,45)

T0= Uniform('T0',0,24)

tau= Gamma('tau',0.0001, 0.0001)

mu_temp= c*T*((T-T0)*(T0<T))*np.sqrt((Tm-T)*(Tm>=T))

mu= 1*(mu_temp>1) + mu_temp*(mu_temp>0)*(mu_temp<1) + 0*(mu_temp<0)

Y_obs = Normal('Y_obs',mu=mu, sd=tau, observed= Y)

# Specifing the parameters that control the MCMC (these will be used throughout the code).

with basic_model_c:

# obtain starting values via MAP

start = find_MAP(fmin=optimize.fmin_powell)

# draw 5000 posterior samples

trace= sample(sample_size, step= Metropolis(), start=start)

#thin the samples by selecting every 5 samples

thin_factor=5

#PLOTTING THE HISTOGRAM

#figure_count= mua.create_2x2_histograms(trace, figure_count)

#Create the Brier Function

Temps= np.arange(0,50,0.1)

c_samps= mua.make_sims_temp_resp("briere_trunc",trace, Temps, thin_factor )

samps['c']=c_samps

#calculate the quants

q= mua.temp_sim_quants(c_samps,Temps)

#PLOTTING THE DATA

#figure_count= mua.add_sim_lines(Temps, c_samps, q,Y , T, figure_count)

#CHECK COUNT:*

#############################################

################# MDR #######################

#############################################

#pEA, b,

data = data_all.query('trait_name== "MDR"')

#GET THE VALUES

Y = data['trait']

Y=Y.as_matrix()

Y= Y.flatten()

#Get the temperatures

T= data['T']

T=T.as_matrix()

T= T.flatten()

size= len(Y)

#TRANSFORM FROM DATAFRAME TO NP ARRAY FORMAT

# Specifing the parameters that control the MCMC (these will be used throughout the code).

basic_model_MDR = Model()

with basic_model_MDR:

#priors for unknown model parameters

c = Gamma('c',1,10)

Tm= Uniform('Tm',25,45)

T0= Uniform('T0',0,24)

tau= Gamma('tau',0.0001, 0.0001)

mu_temp= c*T*((T-T0)*(T0<T))*np.sqrt((Tm-T)*(Tm>T))

mu= 0*(mu_temp<0) + mu_temp*(mu_temp>0)

Y_obs = Normal('Y_obs',mu=mu, sd=tau, observed= Y)

from pymc3 import Metropolis, sample, find_MAP

from scipy import optimize

with basic_model_MDR:

# obtain starting values via MAP

start = find_MAP(fmin=optimize.fmin_powell)

# draw 5000 posterior samples

trace= sample(sample_size, step= Metropolis(), start=start)

#thin the samples by selecting every 5 samples

thin_factor=5

#summary(trace)

#traceplot(trace);

#PLOTTING THE HISTOGRAM

#figure_count= mua.create_2x2_histograms(trace, figure_count)

#Create the Brier Function

Temps= np.arange(0,50,0.1)

MDR_samps= mua.make_sims_temp_resp("briere",trace, Temps, thin_factor )

samps['MDR']=MDR_samps

q= mua.temp_sim_quants(MDR_samps,Temps)

#PLOTTING THE DATA

#figure_count= mua.add_sim_lines(Temps, MDR_samps, q,Y , T, figure_count)

#CHECK COUNT:*

#############################################

################# pEA #######################

#############################################

#pEA, b,

data = data_all.query('trait_name== "pEA"')

#GET THE VALUES

Y = data['trait']

Y=Y.as_matrix()

Y= Y.flatten()

#Get the temperatures

T= data['T']

T=T.as_matrix()

T= T.flatten()

size= len(Y)

#TRANSFORM FROM DATAFRAME TO NP ARRAY FORMAT

# Specifing the parameters that control the MCMC (these will be used throughout the code).

basic_model_pEA = Model()

with basic_model_pEA:

#priors for unknown model parameters

c = Gamma('c',1,1)

Tm= Uniform('Tm',25,45)

T0= Uniform('T0',0,24)

tau= Gamma('tau',0.0001, 0.0001)

mu= -c*((T-T0)*(T0<T))*((T-Tm)*(Tm>T))

Y_obs = Normal('Y_obs',mu=mu, sd=tau, observed= Y)

from pymc3 import Metropolis, sample, find_MAP

from scipy import optimize

with basic_model_pEA:

# obtain starting values via MAP

start = find_MAP(fmin=optimize.fmin_powell)

# draw 5000 posterior samples

trace= sample(sample_size, step= Metropolis(), start=start)

#thin the samples by selecting every 5 samples

thin_factor=5

#summary(trace)

#traceplot(trace);

#PLOTTING THE HISTOGRAM

#figure_count= mua.create_2x2_histograms(trace, figure_count)

#Create the Brier Function

Temps= np.arange(0,50,0.1)

pEA_samps= mua.make_sims_temp_resp("quad_trunc",trace, Temps, thin_factor )

#FIND THE QUANTILES

q= mua.temp_sim_quants(pEA_samps,Temps)

samps['pEA']=pEA_samps

#PLOTTING THE DATA

#figure_count= mua.add_sim_lines(Temps, pEA_samps, q,Y , T, figure_count)

#CHECK COUNT:*

#############################################

################# PDR #######################

#############################################

#pEA, b,

data1 = data_all.query('trait_name== "PDR"')

#GET THE VALUES

Y1 = data1['trait']

Y1=Y1.as_matrix()

Y1= Y1.flatten()

#Get the temperatures

T1= data1['T']

T1=T1.as_matrix()

T1= T1.flatten()

data2 = data_all.query('trait_name== "EIP"')

#GET THE VALUES

Y2 = data2['trait']

Y2=Y2.as_matrix()

Y2= Y2.flatten()

Y2= np.reciprocal(Y2)

#Get the temperatures

T2= data2['T']

T2=T2.as_matrix()

T2= T2.flatten()

#COMBINE

Y= np.append(Y1,Y2)

T= np.append(T1,T2)

size= len(Y)

#TRANSFORM FROM DATAFRAME TO NP ARRAY FORMAT

# Specifing the parameters that control the MCMC (these will be used throughout the code).

basic_model_PDR = Model()

with basic_model_PDR:

#priors for unknown model parameters

c = Gamma('c',1,10)

Tm= Uniform('Tm',25,45)

T0= Uniform('T0',0,24)

tau= Gamma('tau',0.0001, 0.0001)

mu_temp= c*T*((T-T0)*(T0<T))*np.sqrt((Tm-T)*(Tm>T))

mu= 0*(mu_temp<0) + mu_temp*(mu_temp>0)

Y_obs = Normal('Y_obs',mu=mu, sd=tau, observed= Y)

from pymc3 import Metropolis, sample, find_MAP

from scipy import optimize

with basic_model_PDR:

# obtain starting values via MAP

start = find_MAP(fmin=optimize.fmin_powell)

# draw 5000 posterior samples

trace= sample(sample_size, step= Metropolis(), start=start)

#thin the samples by selecting every 5 samples

thin_factor=5

#summary(trace)

#traceplot(trace);

#PLOTTING THE HISTOGRAM

#figure_count= mua.create_2x2_histograms(trace, figure_count)

#Create the Brier Function

Temps= np.arange(0,50,0.1)

PDR_samps= mua.make_sims_temp_resp("briere",trace, Temps, thin_factor )

#FIND THE QUANTILES

q= mua.temp_sim_quants(PDR_samps,Temps)

samps['PDR']=PDR_samps

#PLOTTING THE DATA

#figure_count= mua.add_sim_lines(Temps, PDR_samps, q,Y , T, figure_count)

import pickle

aeg_samps= samps

with open('Aegypti_IPFC_samps.pickle', 'wb') as handle:

pickle.dump(aeg_samps, handle, protocol=pickle.HIGHEST_PROTOCOL)