-
Notifications
You must be signed in to change notification settings - Fork 1
/
iris_training.py
156 lines (134 loc) · 6.21 KB
/
iris_training.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
import os
import sys
import azureml as aml
from azureml.core import Workspace, Datastore, Dataset
from azureml.core.model import Model
from azureml.core.run import Run
import argparse
import json
import time
#import traceback
import logging
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import classification_report, confusion_matrix, precision_score, recall_score, accuracy_score
import pandas as pd
import numpy as np
import re
import math
import seaborn as sn
import matplotlib.pyplot as plt
#from sklearn.externals import joblib
import joblib
'''
IRIS Classification
'''
class IRISClassification():
def __init__(self, args):
'''
Initialize Steps
----------------
1. Initalize Azure ML Run Object
2. Create directories
'''
self.args = args
self.run = Run.get_context()
self.workspace = self.run.experiment.workspace
os.makedirs('./model_metas', exist_ok=True)
def get_files_from_datastore(self, container_name, file_name):
'''
Get the input CSV file from workspace's default data store
Args :
container_name : name of the container to look for input CSV
file_name : input CSV file name inside the container
Returns :
data_ds : Azure ML Dataset object
'''
datastore_paths = [(self.datastore, os.path.join(container_name,file_name))]
data_ds = Dataset.Tabular.from_delimited_files(path=datastore_paths)
dataset_name = self.args.dataset_name
if dataset_name not in self.workspace.datasets:
data_ds = data_ds.register(workspace=self.workspace,
name=dataset_name,
description=self.args.dataset_desc,
tags={'format': 'CSV'},
create_new_version=True)
else:
print('Dataset {} already in workspace '.format(dataset_name))
return data_ds
def create_pipeline(self):
'''
IRIS Data training and Validation
'''
self.datastore = Datastore.get(self.workspace, self.workspace.get_default_datastore().name)
print("Received datastore")
input_ds = self.get_files_from_datastore(self.args.container_name,self.args.input_csv)
final_df = input_ds.to_pandas_dataframe()
print("Input DF Info",final_df.info())
print("Input DF Head",final_df.head())
X = final_df[["SepalLengthCm","SepalWidthCm","PetalLengthCm","PetalWidthCm"]]
y = final_df[["Species"]]
X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.4,random_state=1984)
model = DecisionTreeClassifier()
model.fit(X_train,y_train)
y_pred = model.predict(X_test)
print("Model Score : ", model.score(X_test,y_test))
joblib.dump(model, self.args.model_path)
self.validate(y_test, y_pred, X_test)
match = re.search('([^\/]*)$', self.args.model_path)
# Upload Model to Run artifacts
self.run.upload_file(name=self.args.artifact_loc + match.group(1),
path_or_stream=self.args.model_path)
print("Run Files : ", self.run.get_file_names())
self.run.complete()
def create_confusion_matrix(self, y_true, y_pred, name):
'''
Create confusion matrix
'''
try:
confm = confusion_matrix(y_true, y_pred, labels=np.unique(y_pred))
print("Shape : ", confm.shape)
df_cm = pd.DataFrame(confm, columns=np.unique(y_true), index=np.unique(y_true))
df_cm.index.name = 'Actual'
df_cm.columns.name = 'Predicted'
df_cm.to_csv(name+".csv", index=False)
self.run.upload_file(name="./outputs/"+name+".csv",path_or_stream=name+".csv")
plt.figure(figsize = (120,120))
sn.set(font_scale=1.4)
c_plot = sn.heatmap(df_cm, fmt="d", linewidths=.2, linecolor='black',cmap="Oranges", annot=True,annot_kws={"size": 16})
plt.savefig("./outputs/"+name+".png")
self.run.log_image(name=name, plot=plt)
except Exception as e:
#traceback.print_exc()
logging.error("Create consufion matrix Exception")
def create_outputs(self, y_true, y_pred, X_test, name):
'''
Create prediction results as a CSV
'''
pred_output = {"Actual Species" : y_true['Species'].values, "Predicted Species": y_pred['Species'].values}
pred_df = pd.DataFrame(pred_output)
pred_df = pred_df.reset_index()
X_test = X_test.reset_index()
final_df = pd.concat([X_test, pred_df], axis=1)
final_df.to_csv(name+".csv", index=False)
self.run.upload_file(name="./outputs/"+name+".csv",path_or_stream=name+".csv")
def validate(self, y_true, y_pred, X_test):
self.run.log(name="Precision", value=round(precision_score(y_true, y_pred, average='weighted'), 2))
self.run.log(name="Recall", value=round(recall_score(y_true, y_pred, average='weighted'), 2))
self.run.log(name="Accuracy", value=round(accuracy_score(y_true, y_pred), 2))
self.create_confusion_matrix(y_true, y_pred, "confusion_matrix")
y_pred_df = pd.DataFrame(y_pred, columns = ['Species'])
self.create_outputs(y_true, y_pred_df,X_test, "predictions")
self.run.tag("IRISClassifierFinalRun")
if __name__ == "__main__":
parser = argparse.ArgumentParser(description='QA Code Indexing pipeline')
parser.add_argument('--container_name', type=str, help='Path to default datastore container')
parser.add_argument('--input_csv', type=str, help='Input CSV file')
parser.add_argument('--dataset_name', type=str, help='Dataset name to store in workspace')
parser.add_argument('--dataset_desc', type=str, help='Dataset description')
parser.add_argument('--model_path', type=str, help='Path to store the model')
parser.add_argument('--artifact_loc', type=str,
help='DevOps artifact location to store the model', default='')
args = parser.parse_args()
iris_classifier = IRISClassification(args)
iris_classifier.create_pipeline()