def add_model_to_report(report_dict,modelo):
report_dict[modelo]={glod.get_parameters_key():{},
glod.get_time_parameters_key():{},
glod.get_accuracy_parameter_name():0,
glod.get_decision_index_key():0,
}
return report_dict
def add_model_to_report(report_dict, modelo):
'''This funcion allows to register the information about the trained model in the report'''
report_dict[modelo] = {glod.get_parameters_key():{},
glod.get_time_parameters_key():{},
glod.get_accuracy_parameter_name():0,
glod.get_decision_index_key():0,
}
return report_dict
def create_report_current_model(report_dict,lista_modelos,ruta_relativa_datos_auxiliares,ruta_directorio_informes,enco):
env = Environment(loader=FileSystemLoader('.'))
ruta_plantilla_temporal = os.path.join(ruta_relativa_datos_auxiliares,'temp_html.html')
if(lista_modelos == []): #if process not completed
template = env.get_template(ruta_relativa_datos_auxiliares + '/' + 'incomplete_event_report_template.html') #usamos la plantilla de informes incompletos
template_vars = {glod.get_title_key(): "Incomplete Execution Report",
glod.get_logo_key(): encode_image(report_dict[glod.get_logo_key()].replace('\'','')),
glod.get_report_generic_target_key(): report_dict[glod.get_objective_target_key()],
glod.get_event_key(): report_dict[glod.get_event_key()],
glod.get_info_key(): " " + report_dict['Warning_info']
}
#html
with codecs.open(ruta_plantilla_temporal,'w',encoding=enco) as output_file:
output_file.write(template.render(template_vars))
#pdf
with codecs.open(ruta_plantilla_temporal, 'r') as html_leido:
pdf_resultante=os.path.join(ruta_directorio_informes,"report_"+report_dict[glod.get_event_key()]+"_incomplete.pdf")
with open(pdf_resultante, "wb") as incomplete_rep:
pisa.CreatePDF(html_leido.read(),incomplete_rep)
logging.getLogger("xhtml2pdf").addHandler(PisaNullHandler())
else:
lista_pares_modelo_indice = auxf.order_models_by_score_and_time(report_dict,lista_modelos)
template = env.get_template(ruta_relativa_datos_auxiliares + '/' +'report_template.html') #using standard template
for modelo in lista_modelos:
if(modelo in report_dict):
observations_targets="<p><strong>Target distribution of observations</strong></br>"
for ob_target in auxf.natsorted(report_dict[glod.get_report_general_info_key()][glod.get_report_generic_target_key()].keys()):
observations_targets+=" "+ "With target " + str(ob_target) + " :"+ str(report_dict[glod.get_report_general_info_key()][glod.get_report_generic_target_key()][ob_target]) + "</br>"
observations_targets+="</p>"
variables_summary="<p><strong>Summary of variables</strong></br>"
discarded_for_event = report_dict[glod.get_report_general_info_key()][glod.get_variables_key()][glod.get_user_discarded_key()]
variables_summary+="<br><i><u>Deleted by the user at the begining:</i></u></br>"
for deleted_var in report_dict[glod.get_report_general_info_key()][glod.get_variables_key()][glod.get_deleted_by_user_key()]:
variable_dis=''
if deleted_var in discarded_for_event:
variable_dis = "<strong>" + deleted_var + "</strong>"
else:
variable_dis = deleted_var
variables_summary+=" "+ variable_dis + "</br>"
variables_summary+=" <i>*variables in bold were specified by the user to be discarded specifically for this event<i></br>"
variables_summary+="</br>"
variables_summary+="<br><i><u>Deleted in execution time(Empty or Constant):</i></u></br>"
for emp_con_var in report_dict[glod.get_report_general_info_key()][glod.get_variables_key()][glod.get_empty_or_constant_key()]:
variables_summary+=" "+ emp_con_var + "</br>"
variables_summary+="</br>"
variables_summary+="<br><i><u>Requested for the event by the user:</i></u></br>"
for req_var in report_dict[glod.get_report_general_info_key()][glod.get_variables_key()][glod.get_user_requested_key()]:
variables_summary+=" "+ req_var + "</br>"
variables_summary+="</br>"
variables_summary+="<br><i><u>Used during the process:</i></u></br>"
diccionario_relevantes_mif = report_dict[glod.get_report_general_info_key()][glod.get_variables_key()][glod.get_score_relevant_key()]
sorted_relevant_vars = sorted(diccionario_relevantes_mif.items(), key=operator.itemgetter(1), reverse=True)
for relevant_var in sorted_relevant_vars:
rel_variable= relevant_var[0]
rel_variable = "<strong>" + rel_variable +' '+ str(diccionario_relevantes_mif[rel_variable]) +"</strong>"
variables_summary+=" "+ rel_variable + "</br>"
for relevant_var in report_dict[glod.get_report_general_info_key()][glod.get_variables_key()][glod.get_used_in_process()]:
if (relevant_var not in diccionario_relevantes_mif) :
variables_summary+=" "+ relevant_var + "</br>"
variables_summary+=" <i>*variables in bold were used to train the models<i></br>"
variables_summary+="</p>"
#Information about the model
accuracy = "</br></br> <strong>Accuracy: "+ str(float(round(report_dict[modelo][glod.get_accuracy_parameter_name()],5)))+"</strong>"
ranking = get_string_with_ranking_of_models(lista_pares_modelo_indice,modelo)
model_info = "<p><strong>Parameters used to configure the model</strong></br>"
for param in report_dict[modelo][glod.get_parameters_key()]:
model_info+= " <i>"+ param + "</i>: " + str(report_dict[modelo][glod.get_parameters_key()][param]) + "</br>"
model_info+="</p>"
time_info = "<p><strong>Time elapsed</strong></br>"
tiempo_seleccion_parametros = report_dict[modelo][glod.get_time_parameters_key()][glod.get_time_sel_finish_key()] - report_dict[modelo][glod.get_time_parameters_key()][glod.get_time_sel_init_key()]
tiempo_entrenamiento = report_dict[modelo][glod.get_time_parameters_key()][glod.get_time_train_finish_key()] - report_dict[modelo][glod.get_time_parameters_key()][glod.get_time_train_init_key()]
time_info+=" "+ "Parameters selection time: "+ str(tiempo_seleccion_parametros) + "</br>"
time_info+=" "+ "Training time: "+ str(tiempo_entrenamiento) + "</br>"
time_info+="</p>"
total_train = 0.0
vector_of_targets = []
vector_of_values_by_target = []
vector_of_percentages_by_target = []
train_distribution_info ="<p></br><strong>Training Data Distribution</strong></br>"
for train_target in auxf.natsorted(report_dict[glod.get_report_general_info_key()][glod.get_training_division_key()].keys()):
train_distribution_info+=" "+ "With target " + str(train_target) + " :"+ str(report_dict[glod.get_report_general_info_key()][glod.get_training_division_key()][train_target]) + "</br>"
vector_of_targets.append(train_target)
vector_of_values_by_target.append(float(report_dict[glod.get_report_general_info_key()][glod.get_training_division_key()][train_target]))
total_train+=float(report_dict[glod.get_report_general_info_key()][glod.get_training_division_key()][train_target])
train_distribution_info+="</p>"
#getting null train accuracy
null_train_accuracy = 0.0
for indice_t in range(len(vector_of_values_by_target)):
vector_of_percentages_by_target.append(round(vector_of_values_by_target[indice_t]/total_train,4))
null_train_accuracy = max(vector_of_percentages_by_target)
total_test = 0.0
vector_of_targets = []
vector_of_values_by_target = []
vector_of_percentages_by_target = []
test_distribution_info ="<p><strong>Test Data Distribution</strong></br>"
for test_target in auxf.natsorted(report_dict[glod.get_report_general_info_key()][glod.get_test_division_key()].keys()):
test_distribution_info+=" "+ "With target " + str(test_target) + " :"+ str(report_dict[glod.get_report_general_info_key()][glod.get_test_division_key()][test_target]) + "</br>"
vector_of_targets.append(test_target)
vector_of_values_by_target.append(float(report_dict[glod.get_report_general_info_key()][glod.get_test_division_key()][test_target]))
total_test+=float(report_dict[glod.get_report_general_info_key()][glod.get_test_division_key()][test_target])
test_distribution_info+="</p>"
null_test_accuracy = 0.0
for indice_t in range(len(vector_of_values_by_target)):
vector_of_percentages_by_target.append(round(vector_of_values_by_target[indice_t]/total_test,4))
null_test_accuracy = max(vector_of_percentages_by_target)
event = report_dict[glod.get_event_key()]
template_vars = {glod.get_title_key(): "Execution Report",
glod.get_logo_key():encode_image(report_dict[glod.get_logo_key()].replace('\'','')),
glod.get_model_key(): modelo,
glod.get_report_generic_target_key(): report_dict[glod.get_objective_target_key()],
glod.get_event_key(): event,
glod.get_accuracy_parameter_name(): str(accuracy)+"<br> <strong>Null train acc: "+ str(null_train_accuracy)+"</strong>"+"<br> <strong>Null test acc: "+ str(null_test_accuracy)+ "</strong></p>",
glod.get_models_ranking_key(): ranking,
glod.get_observations_targets_key(): observations_targets,
glod.get_variables_summary_key(): variables_summary,
glod.get_models_info_key(): model_info,
glod.get_time_info_key(): time_info,
glod.get_train_distribution_info_key(): train_distribution_info,
glod.get_test_distribution_info_key(): test_distribution_info
}
template_vars[glod.get_metrics_info_key()] =""
for metric in report_dict[modelo][glod.get_metrics_micro_avg_key()]:
template_vars[glod.get_metrics_info_key()] +="<p>"+"<strong>"+metric+"</strong>: " + report_dict[modelo][glod.get_metrics_micro_avg_key()][metric] +"</br>"
template_vars[glod.get_metrics_info_key()] +="</p>"
if glod.get_model_parameters_plot_name() in report_dict[modelo]:
template_vars[glod.get_image_parameters_accuracy_key()] = encode_image(report_dict[modelo][glod.get_model_parameters_plot_name()].replace('\'',''))
if glod.get_confussion_matrix_train_path_key() in report_dict[modelo]:
template_vars[glod.get_conf_train_img_key()] = encode_image(report_dict[modelo][glod.get_confussion_matrix_train_path_key()].replace('\'',''))
if glod.get_confussion_matrix_test_path_key() in report_dict[modelo]:
template_vars[glod.get_conf_test_img_key()] = encode_image(report_dict[modelo][glod.get_confussion_matrix_test_path_key()].replace('\'',''))
if(glod.get_learning_curve_key() in report_dict[modelo]):
template_vars[glod.get_learning_curve_key()] = encode_image(report_dict[modelo][glod.get_learning_curve_key()].replace('\'',''))
metrics_by_label = "<table width='100%' border='1' cellspacing='0' cellpadding='5'>"
keys = ''
for elemento in auxf.natsorted(report_dict[modelo][glod.get_metrics_key()].keys()):
if(keys == ''):
keys = report_dict[modelo][glod.get_metrics_key()][elemento].keys()
metrics_by_label+="<tr><td align='center' class='black'>"+ glod.get_report_generic_target_key() +"</td>"
for cabecera in keys:
metrics_by_label+="<td align='center' class='black'>" + cabecera +"</td>"
metrics_by_label += "</tr>"
metrics_by_label+= "<tr><td>" + elemento.replace('target_','') + "</td>"
for key in keys:
metrics_by_label += "<td>"+str(report_dict[modelo][glod.get_metrics_key()][elemento][key])+"</td>"
metrics_by_label+= "</tr>"
metrics_by_label+="</table>"
template_vars[glod.get_metrics_by_label_key()] = metrics_by_label
#generamos el html
with codecs.open(ruta_plantilla_temporal,'w',encoding=enco) as output_file:
output_file.write(template.render(template_vars))
#generamos el pdf
with codecs.open(ruta_plantilla_temporal, mode='r',encoding=enco) as read_html:
pdf_resultante=os.path.join(ruta_directorio_informes,modelo + "_report_for_"+ event +".pdf")
with open(pdf_resultante, mode='wb') as pdf_gen:
pisa.CreatePDF(read_html.read(),pdf_gen)
logging.getLogger("xhtml2pdf").addHandler(PisaNullHandler())
if(os.path.exists(ruta_plantilla_temporal)):
os.remove(ruta_plantilla_temporal)
def update_model_parameters(report_dict,modelo,parametro,valor):
report_dict[modelo][glod.get_parameters_key()][parametro] = valor
return report_dict
def update_model_parameters(report_dict, modelo, parametro, valor):
'''This funcion allows to register the information about the parameters of the
trained model in the report'''
report_dict[modelo][glod.get_parameters_key()][parametro] = valor
return report_dict