logger.info('Updated image generation parameters: %s', image_generation_params)
#Compute predictions
num_prediction_steps = num_prediction_steps or ceil(len(input_data) / image_generation_params.batch_size)
predictor = Prediction(model, input_params, image_generation_params)
predicted_data = predictor.predict(input_data, num_prediction_steps)
#Compute accuracy
num_matches = (predicted_data[constants.PANDAS_MATCH_COLUMN].to_numpy().nonzero())[0].shape[0]
num_mismatches = len(predicted_data[constants.PANDAS_MATCH_COLUMN]) - num_matches
accuracy = (num_matches/len(predicted_data[constants.PANDAS_MATCH_COLUMN])) * 100.
#Write-out predicted output
prediction_result_file = InputDataFile(constants.PREDICTION_RESULT_FILE_NAME_GUIDANCE)
prediction_result_file.save(predicted_data, 0, 0)
input_files_client.put_all([prediction_result_file.file_name(0, 0)])
print_summary = """
Result Dataframe: {}
Total predictions: {}
Correct predictions: {}
Wrong predictions: {}
Accuracy: {}
""".format(
predicted_data,
len(predicted_data),
num_matches,
num_mismatches,
accuracy)
#Print summary
#Prepare input files
input_files_client = InputFiles(dropbox)
input_data = input_files_client.get_all([input_data])[input_data]
#Input data as pandas data frame
input_data = csv_to_dataframe(input_data)
####################################### Prepare the input dataset [End] ############################################
####################################### Rebalance the dataset [Start] ############################################
#Rebalance the data and obtain the statistics
rebalancer = Rebalancing(input_data, label_col)
result, pre_stats, post_stats = rebalancer.rebalance(statistics=True)
####################################### Rebalance the dataset [End] ############################################
#Output to a file
dataframe_to_csv(result, output_file)
####################################### Process output [Start] ############################################
#Upload the output to dropbox
if dropbox:
input_files_client.put_all([output_file])
print_summary = """
Pre-balancing statistics: {}
Results: {}
Post-balancing statistics: {}
""".format(pre_stats, result, post_stats)
print(print_summary)
####################################### Process output [End] ############################################