ScoringScript.txt

# coding: utf-8

# In[ ]:

import sys
import numpy as np
import pandas as pd
import csv
import pickle


# reading attribute file as a dictionary
with open(sys.argv[1], mode='r') as infile:
    reader = csv.reader(infile)
    with open(sys.argv[1] + '_Dummy.csv', mode='w') as outfile:
        writer = csv.writer(outfile)
        attr_dict = {rows[0]:rows[1] for rows in reader}

# reading variable list file into a list
with open(attr_dict['VariableList'], 'r') as f:
    reader = csv.reader(f)
    variable_list = list(reader)

variable_list = variable_list[0]


# loading the imputer, scaler, pca nd xgboost model
imputer = pickle.load(open(attr_dict['Imputer'], "rb"))
scaler = pickle.load(open(attr_dict['Scaler'], "rb"))
pca = pickle.load(open(attr_dict['PCA'], "rb"))

model = pickle.load(open(attr_dict['Model'], "rb"))


#scoring the data
data_to_score = pd.read_csv(attr_dict['RolledUpData'])

prediction =  model.predict(pca.transform(scaler.transform(imputer.transform(data_to_score[variable_list]))))
prediction = prediction/100

boundaries = [0] + list(np.arange(0.06, 0.301, 0.005)) + [0.31, 0.32, 0.33, 0.34, 0.35, np.inf]
bucket = pd.cut(list(prediction), boundaries).astype(str)[0]

with open(attr_dict['ActualP50'], mode='r') as infile:
    reader = csv.reader(infile)
    with open(attr_dict['ActualP50'] + '_Dummy.csv', mode='w') as outfile:
        writer = csv.writer(outfile)
        predicted_buckets = {rows[0]:rows[1] for rows in reader}

Actual_P50 = predicted_buckets[bucket]
print(Actual_P50)