-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathPDF_ReportApplyJob.py
240 lines (206 loc) · 11.1 KB
/
PDF_ReportApplyJob.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
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
"""
File: PDF_ReportApplyJob.py
Authors: Ryan J. Urbanowicz, Richard Zhang, Wilson Zhang
Institution: University of Pensylvania, Philadelphia PA
Creation Date: 6/1/2021
License: GPL 3.0
Description: Phase 10 of AutoMLPipe-BC - This 'Job' script is called by PDF_ReportApplyMain.py which generates a formatted PDF summary report of key
pipeline results (applying trained models to hold out replication data). It is run once.
"""
#Import required packages ---------------------------------------------------------------------------------------------------------------------------
import pandas as pd
from fpdf import FPDF
from datetime import datetime
import glob
import os
import re
import sys
def job(experiment_path,rep_data_path,data_path):
time = str(datetime.now())
train_name = data_path.split('/')[-1].split('.')[0]
experiment_name = experiment_path.split('/')[-1]
#Function to Convert Dataset lists into Usable Strings to Write to the PDF
#Find folders inside directory
ds = []
for datasetFilename in glob.glob(rep_data_path+'/*'):
datasetFilename = str(datasetFilename).replace('\\','/')
apply_name = datasetFilename.split('/')[-1].split('.')[0] #Save unique dataset names so that analysis is run only once if there is both a .txt and .csv version of dataset with same name.
ds.append(apply_name)
ds = sorted(ds)
print(ds)
ars_df = pd.read_csv(experiment_path+ '/'+'metadata.csv')
ars_dic = []
for i in range(len(ars_df)):
if i >= 0:
ars_dic.append(ars_df.iloc[i, 0]+': ')
ars_dic.append(ars_df.iloc[i, 1])
ars_dic.append('\n')
else:
pass
#Analysis Settings, Global Analysis Settings, ML Modeling Algorithms
analy_report = FPDF('P', 'mm', 'A4')
analy_report.set_margins(left=10, top=5, right=10, )
analy_report.add_page(orientation='P')
top = analy_report.y
#ML Pipeline Analysis Report-------------------------------------------------------------------------------------------------------
print("Starting Report")
ls1 = ars_dic[0:59] # Class - filter poor [0:55] 59
ls2 = ars_dic[59:98] #ML modeling algorithms (NaiveB - ExSTraCS) [56:95] 60
ls3 = ars_dic[98:114] #primary metric - hypersweep timeout [94:111] 97
ls4 = ars_dic[114:129] #LCS parameters (do LCS sweep - LCS hypersweep timeout) [110:125]
analy_report.set_font('Times', 'B', 12)
analy_report.cell(w=180, h=8, txt='AutoMLPipe-BC Apply Summary Report: '+time, ln=2, border=1, align='L')
analy_report.y += 3
analy_report.set_font(family='times', size=9)
analy_report.multi_cell(w = 90,h = 4,txt='Pipeline Settings:'+'\n'+'\n'+listToString(ls1)+' '+listToString(ls3), border=1, align='L')
analy_report.x += 90
analy_report.y = analy_report.y - 104 #96
analy_report.multi_cell(w = 90,h = 4,txt='ML Modeling Algorithms:'+'\n'+'\n'+listToString(ls2), border=1, align='L')
analy_report.x += 90
analy_report.y += 4
analy_report.multi_cell(w = 90,h = 4,txt='LCS Settings (ExSTraCS,eLCS,XCS):'+'\n'+listToString(ls4), border=1, align='L')
analy_report.y +=15 #10
analy_report.cell(w = 180, h = 4, txt='Target Training Dataset: '+train_name, border=1, align='L')
analy_report.y +=8
analy_report.x = 10
listDatasets = ''
i = 1
for each in ds:
listDatasets = listDatasets+('D'+str(i)+' = '+str(each)+'\n')
i += 1
analy_report.multi_cell(w = 180, h = 4, txt='Applied Datasets: '+'\n'+listDatasets, border=1, align='L')
footer(analy_report)
#Data and Model Prediction Summary--------------------------------------------------------------------------------------
print("Publishing Model Prediction Summary")
for n in range(len(ds)):
#Create PDF and Set Options
analy_report.set_margins(left=1, top=1, right=1, )
analy_report.add_page()
analy_report.set_font('Times', 'B', 12)
analy_report.cell(w=0, h = 8, txt="Dataset and Model Prediction Summary: D"+str(n+1)+" = "+ds[n], border=1, align="L", ln=2)
analy_report.set_font(family='times', size=9)
#Exploratory Analysis ----------------------------
analy_report.image(experiment_path+'/'+train_name+'/applymodel/'+ds[n]+'/exploratory/ClassCountsBarPlot.png', 5, 10, 70,48) #10, 30, 82)
analy_report.x = 125
analy_report.y = 55
try:
analy_report.image(experiment_path+'/'+train_name+'/applymodel/'+ds[n]+'/exploratory/FeatureCorrelations.png', 85, 10, 115) #88, 30, 120, 60)
except:
analy_report.cell(40, 4, 'No Feature Correlation Plot', 1, align="L")
pass
data_summary = pd.read_csv(experiment_path+'/'+train_name+'/applymodel/'+ds[n]+"/exploratory/DataCounts.csv")
info_ls = []
for i in range(len(data_summary)):
info_ls.append(data_summary.iloc[i,0]+': ')
info_ls.append(str(data_summary.iloc[i,1]))
info_ls.append('\n')
analy_report.x = 5
analy_report.y = 58
analy_report.multi_cell(w=60, h=4, txt='Variable: Count'+'\n'+listToString(info_ls), border=1, align='L')
#Report Best Algorithms by metric
summary_performance = pd.read_csv(experiment_path+'/'+train_name+'/applymodel/'+ds[n]+"/model_evaluation/Summary_performance_mean.csv")
summary_performance['ROC_AUC'] = summary_performance['ROC_AUC'].astype(float)
highest_ROC = summary_performance['ROC_AUC'].max()
algorithm = summary_performance[summary_performance['ROC_AUC'] == highest_ROC].index.values
best_alg_ROC = summary_performance.iloc[algorithm, 0]
summary_performance['Balanced Accuracy'] = summary_performance['Balanced Accuracy'].astype(float)
highest_BA = summary_performance['Balanced Accuracy'].max()
algorithm = summary_performance[summary_performance['Balanced Accuracy'] == highest_BA].index.values
best_alg_BA = summary_performance.iloc[algorithm, 0]
summary_performance['F1_Score'] = summary_performance['F1_Score'].astype(float)
highest_F1 = summary_performance['F1_Score'].max()
algorithm = summary_performance[summary_performance['F1_Score'] == highest_F1].index.values
best_alg_F1 = summary_performance.iloc[algorithm, 0]
summary_performance['PRC_AUC'] = summary_performance['PRC_AUC'].astype(float)
highest_PRC = summary_performance['PRC_AUC'].max()
algorithm = summary_performance[summary_performance['PRC_AUC'] == highest_PRC].index.values
best_alg_PRC = summary_performance.iloc[algorithm, 0]
summary_performance['PRC_APS'] = summary_performance['PRC_APS'].astype(float)
highest_APS = summary_performance['PRC_APS'].max()
algorithm = summary_performance[summary_performance['PRC_APS'] == highest_APS].index.values
best_alg_APS = summary_performance.iloc[algorithm, 0]
analy_report.x = 5
analy_report.y = 87
analy_report.multi_cell(w=70, h=4, txt="Best (ROC_AUC): "+ str(best_alg_ROC.values)+' = '+ str("{:.3f}".format(highest_ROC))+
'\n'+"Best (Balanced Acc.): "+ str(best_alg_BA.values)+' = '+ str("{:.3f}".format(highest_BA))+
'\n'+"Best (F1 Score): "+ str(best_alg_F1.values)+' = '+ str("{:.3f}".format(highest_F1))+
'\n'+"Best (PRC_AUC): "+ str(best_alg_PRC.values)+' = '+ str("{:.3f}".format(highest_PRC))+
'\n'+"Best (PRC_APS): "+ str(best_alg_APS.values)+' = '+ str("{:.3f}".format(highest_APS)), border=1, align='L')
#ROC-------------------------------
analy_report.x = 5
analy_report.y = 112
analy_report.cell(10, 4, 'ROC', 1, align="L")
analy_report.image(experiment_path+'/'+train_name+'/applymodel/'+ds[n]+'/model_evaluation/Summary_ROC.png', 4, 118, 120)
analy_report.image(experiment_path+'/'+train_name+'/applymodel/'+ds[n]+'/model_evaluation/metricBoxplots/Compare_ROC_AUC.png', 124, 118, 82,85)
#PRC-------------------------------
analy_report.x = 5
analy_report.y = 200
analy_report.cell(10, 4, 'PRC', 1, align="L")
analy_report.image(experiment_path+'/'+train_name+'/applymodel/'+ds[n]+'/model_evaluation/Summary_PRC.png', 4, 206, 133) #wider to account for more text
analy_report.image(experiment_path+'/'+train_name+'/applymodel/'+ds[n]+'/model_evaluation/metricBoxplots/Compare_PRC_AUC.png', 138, 205, 68,80)
footer(analy_report)
#Average Model Prediction Statistics--------------------------------------------------------------------------------------
print("Publishing Average Model Prediction Statistics")
for n in range(len(ds)):
#Create PDF and Set Options
analy_report.set_margins(left=1, top=1, right=1, )
analy_report.add_page()
analy_report.set_font('Times', 'B', 12)
analy_report.cell(w=0, h = 8, txt="Average Model Prediction Statistics: D"+str(n+1)+" = "+ds[n], border=1, align="L", ln=2)
analy_report.set_font(family='times', size=7)
stats_ds = pd.read_csv(experiment_path+'/'+train_name+'/applymodel/'+ds[n]+'/model_evaluation/Summary_performance_mean.csv',sep=',',index_col=0)
stats_ds = stats_ds.round(4)
#Format
stats_ds.reset_index(inplace=True)
stats_ds = stats_ds.columns.to_frame().T.append(stats_ds, ignore_index=True)
stats_ds.columns = range(len(stats_ds.columns))
epw = 208 #Amount of Space (width) Avaliable
th = analy_report.font_size
col_width = epw/float(10) #maximum column width
#Print next 3 datasets
table1 = stats_ds.iloc[: , :10]
table1 = table1.to_numpy()
for row in table1:
for datum in row:
analy_report.cell(col_width, th, str(datum), border=1)
analy_report.ln(th) #critical
analy_report.y += 5
table1 = stats_ds.iloc[: , 10:18]
met = stats_ds.iloc[:,0]
table1 = pd.concat([met, table1], axis=1)
table1 = table1.to_numpy()
for row in table1:
for datum in row:
analy_report.cell(col_width, th, str(datum), border=1)
analy_report.ln(th) #critical
analy_report.y += 5
footer(analy_report)
#Output The PDF Object
try:
fileName = str(experiment_name)+'_ML_Pipeline_Apply_Report.pdf'
analy_report.output(experiment_path+'/'+train_name+'/applymodel/'+ds[n]+'/'+fileName)
print('PDF Generation Complete')
except:
print('Pdf Output Failed')
def listToString(s):
str1 = " "
return (str1.join(s))
#Create Footer
def footer(self):
self.set_auto_page_break(auto=False, margin=3)
self.set_y(285)
self.set_font('Times', 'I', 7)
self.cell(0, 7,'Generated with the URBS-Lab AutoMLPipe-BC: (https://github.com/UrbsLab/AutoMLPipe-BC)', 0, 0, 'C')
self.set_font(family='times', size=9)
#Find N greatest ingegers within a list
def ngi(list1, N):
final_list = []
for i in range(0, N):
max1 = 0
for j in range(len(list1)):
if list1[j] > max1:
max1 = list1[j];
list1.remove(max1);
final_list.append(max1)
if __name__ == '__main__':
job(sys.argv[1],sys.argv[2],sys.argv[3])