def run(self):
import rfprediction
fn_t = self.parameters['filename_training']
fn_i = self.parameters['filename_input']
dir_out = self.parameters['directory_report']
output_column = self.parameters.get('output_field', 'OUT')
id_column = self.parameters.get('id_field', 'ID')
num_trees = self.parameters.get('num_trees', 20)
max_depth = self.parameters.get('max_depth', 4)
iterations = self.parameters.get('iterations', 1)
# if isinstance(iterations, int) is False: iterations = 1
# if isinstance(num_trees, int) is False: num_trees = 20
# if isinstance(max_depth, int) is False: max_depth = 4
try:
summary = rfprediction.execute_analysis(training_file=fn_t,
diagnosis_file=fn_i,
mum_trees=num_trees,
max_depth=max_depth,
id_field=id_column,
output_field=output_column,
output=dir_out,
iterations=iterations)
# result = report_file
# summary = rfprediction.execute_analysis()
error = None
import rfreport
result = rfreport.generate_report_document(summary, dir_out)
except Exception as e:
result = None
error = repr(e)
event = MainPanel._event_class(error=error, result=result)
wx.PostEvent(self.parent, event) # throw results via messaging system
try:
predicted = get_decision_results(best_forest, predictionset, fields)
except Exception as e:
sys.stderr.write('error while prediction : ' + e)
raise e
# Save data
summary = pack_json_results(trainingset, predictionset, fields, predicted, best_forest, best_tree, output_groups, weights, conditions)
if args.key: # unique key for interaction with other processes
summary['key'] = args.key
if args.without_rawdata: # remove rawdata for privacy concern
fields = [KEYWORD_ID, KEYWORD_OUTPUT]
summary['trainingset'] = __trim_data_fields(summary['trainingset'], fields)
summary['analysisset'] = __trim_data_fields(summary['trainingset'], fields)
if dstdir.lower().endswith('.json'): # save JSON only
filename = dstdir
dstdir = None
with open(filename, 'w') as fo:
json.dump(summary, fo, indent=4, separators=(',', ': '))
else: # HTML reports
if os.path.exists(dstdir) is False:
os.makedirs(dstdir)
filename = os.path.join(dstdir, 'report_{}.json'.format(timestamp))
with open(filename, 'w') as fo:
json.dump(summary, fo, indent=4, separators=(',', ': '))
# visualization
import rfreport
rfreport.generate_report_document(summary, dstdir, timestamp=timestamp)
conditions = {}
conditions['Training data'] = args.t
conditions['Analysis data'] = args.i
conditions['Number of Trees'] = args.n
conditions['Maximum depth'] = args.d
conditions['Iteration'] = args.iteration
if args.verbose:
for key, value in conditions.items():
sys.stderr.write('{}\t{}\n'.format(key, value))
filename_prediction = args.i if args.i is not None else args.t
iteration = args.iteration
dstdir = args.o
trainingset, predictionset, fields \
= load_files_and_determine_fields(filename_training=args.t, filename_diagnosis=args.i, field_id=args.I, field_output=args.F, verbose=args.verbose)
best_forest, best_tree, weights, output_groups = _obtain_forest(trainingset, predictionset, fields, args.n, args.d, iteration, args.verbose)
if predictionset is None:
predictionset = copy.deepcopy(trainingset)
predicted = get_decision_results(best_forest, predictionset, fields)
timestamp = __get_timestamp()
filename = os.path.join(dstdir, 'report_{}.json'.format(timestamp))
summary = save_json_results(filename, trainingset, predictionset, fields, predicted, best_forest, best_tree, output_groups, weights, conditions)
import rfreport
rfreport.generate_report_document(summary, dstdir)
# execute_analysis(training_file=args.t, diagnosis_file=args.i, num_trees=args.n,
# max_depth=args.d, output=args.o, output_field=args.F, id_field=args.I, verbose=args.verbose)
