-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathStatFuncs.py
224 lines (192 loc) · 8.61 KB
/
StatFuncs.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
from logging import critical
from os import stat
import pandas as pd
from flask import jsonify
from sklearn.neighbors import NearestNeighbors
import numpy as np
from pycaret.regression import *
sr_sys_counts=pd.read_csv("./public/data/sr_sys_counts.csv")
ib = pd.read_csv("./public/data/Hackathon_IB_Data_1.csv")
sr = pd.read_csv("./public/data/Hackathon_SR_Data_1.csv")
ec = pd.read_csv("./public/data/Hackathon_exam_count_Data.csv")
status_info=pd.read_csv("./public/data/all_data.csv",index_col=False)
ib['installdate'] = pd.to_datetime(ib['installdate'],errors = 'coerce').dt.normalize()
ib['deinstalldate'] = pd.to_datetime(ib['deinstalldate'],errors = 'coerce').dt.normalize()
sr['sr_open_date'] = pd.to_datetime(sr['sr_open_date'],errors = 'coerce').dt.normalize()
sr['sr_close_date'] = pd.to_datetime(sr['sr_close_date'],errors = 'coerce').dt.normalize()
def to_json(data):
result_data = []
row_count = data.shape[0]
column_count = data.shape[1]
column_names = data.columns.tolist()
final_row_data = []
for index, rows in data.iterrows():
final_row_data.append(rows.to_dict())
json_result = {'rows': row_count, 'cols': column_count, 'columns': column_names, 'rowData': final_row_data}
result_data.append(json_result)
return jsonify(result_data)
def toSeries(df):
#row_data=[list(df.columns)]
col = list(df.columns)
row_data = []
for index, rows in df.iterrows():
row_data.append(list(rows))
json_res = {"rows" : len(row_data), "columns" : len(col), "row_data" : row_data, "col_data" : col }
return json_res
def get_sr_costs():
src=sr.groupby('dummy_sysid').sum('Cos').drop('hour',axis='columns')
srh=sr.groupby('dummy_sysid').sum('hour').drop('Cos',axis='columns')
return {'cost':src,'hours':srh}
def get_sys():
ib.drop_duplicates(inplace=True)
ib.drop(['ownershiptype','systemcoveragelevelwarrantyc','transferacceptancedate','shippeddate','last_covered_date'],axis='columns',inplace=True)
return toSeries(ib)
def get_service_plot():
# returns service counts for all the sys ids
counts=sr["dummy_sysid"].groupby(sr.sr_open_date.dt.year).count().to_frame().reset_index()
#return counts.to_dict()
return toSeries(counts)
def get_service_month():
counts=sr["dummy_sysid"].groupby(sr.sr_open_date.dt.month).count().to_frame().reset_index()
return toSeries(counts)
def get_parts_counts():
counts=sr["dummy_sysid"].groupby(sr.dummy_part_number).count().to_frame().reset_index()
counts=counts[counts['dummy_sysid']>=10]
#counts = counts.sort_values('dummy_sysid',ascending=True)
return toSeries(counts)
def get_freq_sys():
freqs=sr_sys_counts.loc[:,["dummy_sysid","count"]]
freqs=freqs.sort_values('count',ascending=True).head(50)
return toSeries(freqs)
def get_ec_stats():
stats = ec.groupby(['aggr_month','aggr_year']).agg({'aggr_value':'sum'}).reset_index()
stats = stats.sort_values(['aggr_year', 'aggr_month'], ascending=True)
return toSeries(stats)
def get_total_ec():
counts = ec.groupby(['dummy_sysid']).agg({'aggr_value':'sum'}).reset_index()
counts = counts.sort_values('aggr_value', ascending=False)
return toSeries(counts)
def ec_max_month():
ec1 = ec.dropna(subset=["dummy_sysid"])
stats=ec1.loc[ec1.groupby(['aggr_month','aggr_year'])['aggr_value'].idxmax()]
return toSeries(stats[["dummy_sysid","aggr_value"]])
def get_labels():
gc = status_info.groupby(['Label']).count()
return gc['dummy_sysid'].to_dict()
def get_devices():
cr=status_info
return toSeries(cr)
def predict(sysid):
label = status_info.loc[status_info['dummy_sysid']==sysid].drop(['dummy_part_number'],axis='columns')
return toSeries(label)
# Part specific Functions :
def avgTimeBetweenServices(sysid):
sr_dates = sr[sr['dummy_sysid']==sysid]
sr_dates = sr_dates[['sr_open_date', 'sr_close_date']].reset_index()
sum = 0
for i in range(1, len(sr_dates)) :
sum = sum + ((abs(sr_dates.loc[i, "sr_open_date"] - sr_dates.loc[i-1, "sr_close_date"])).days)
return(sum/(len(sr_dates)-1))
def avgDownTime(sysid):
n = sr[sr['dummy_sysid']==sysid][['sr_open_date', 'sr_close_date']]
return((n['sr_close_date'] - n['sr_open_date']).mean().days)
def avgSRCount(sysid):
yr = list(pd.DatetimeIndex(sr[sr['dummy_sysid']==sysid]['sr_open_date']).year)
sum = 0
for i in set(yr):
sum = sum+yr.count(i)
return(sum/(len(set(yr))))
def get_num_replaced(sysid):
count_replaced = sr.loc[sr['dummy_sysid']==sysid]['dummy_part_number'].count()
return count_replaced
def get_sys_sr(sysid):
first_sr = (sr_sys_counts.loc[sr_sys_counts['dummy_sysid']==sysid]).sort_values('sr_open_date').reset_index()['sr_open_date'][0]
return first_sr
def get_sys_install_date(sysid):
install_date = ib.loc[ib['dummy_sysid']==sysid]['installdate'].unique()[0]
return pd.to_datetime(install_date)
# Get similar systems to queried system
def find_nearest_system(sysid):
all_data = pd.read_csv("./public/data/all_data_no_label.csv",index_col=False)
#all_data.drop(columns=["Label"])
ad2=all_data.dropna()
idx = ad2.loc[ad2['dummy_sysid']==sysid]
query_vector = idx.to_numpy()
query_vector = np.delete(query_vector, [1])
ad2.drop(columns=["dummy_sysid"],inplace=True)
vector = ad2.to_numpy()
neigh = NearestNeighbors(n_neighbors=3)
neigh.fit(ad2)
neighbors = neigh.kneighbors([query_vector], return_distance = False)
return list(all_data.iloc[neighbors[0][1:]]['dummy_sysid'])
def check_in_service(sysid):
sr_sort = sr.sort_values(by=["sr_open_date"])
sr_sort.drop_duplicates(inplace=True)
sys_row=sr_sort[sr_sort["dummy_sysid"]==sysid].iloc[-1]
sr_diff = (pd.Timestamp.now().normalize()-sys_row['sr_close_date'])/ np.timedelta64(1, 'D')
if sr_diff>0:
return "System currently in service"
else:
return "System currently not in service"
def get_device_stats(sysid):
d= {}
d['bw_sr'] = avgTimeBetweenServices(sysid)
d['down'] = avgDownTime(sysid)
d['sr_count'] = avgSRCount(sysid)
d['parts_replaced'] = get_num_replaced(sysid)
d['install_date'] = get_sys_install_date(sysid)
d['first_sr'] = get_sys_sr(sysid)
d['neareast_neigh'] = [find_nearest_system(sysid)]
return toSeries(pd.DataFrame(d, index=[0]))
#print(get_device_stats('sys1018'))
print(find_nearest_system('sys1018'))
def get_exam_counts(sysid):
sys_ec = ec[ec['dummy_sysid']==sysid]
return sys_ec.shape[0]
def get_sr_counts(sysid):
sys_sr = sr[sr['dummy_sysid']==sysid]
return sys_sr.shape[0]
def get_age(sysid):
sys = ib[ib["dummy_sysid"]==sysid]
age=(pd.Timestamp.now().normalize()-sys['installdate'])/ np.timedelta64(1, 'Y')
try:
return list(age)[0]
except IndexError:
return 0
def get_last_service(sysid):
sys = sr[sr["dummy_sysid"]==sysid]
last_service=(pd.Timestamp.now().normalize()-sys['sr_close_date'])/ np.timedelta64(1, 'Y')
try:
return sorted(list(last_service),reverse=True)[0]
except IndexError:
return 0
def get_sys_parts(sysid):
lgbm=load_model('PartModel')
sys_parts = sr[sr['dummy_sysid']==sysid]
sys_parts.dropna(subset=["dummy_part_number"],inplace=True)
sr_count = get_sr_counts(sysid)
ec_count=get_exam_counts(sysid)
age = get_age(sysid)
last_sr= get_last_service(sysid)
sys_parts["sr_freq"]=sys_parts["dummy_sysid"].apply(get_sr_counts)
sys_parts["ageagainstinstallation"]=[age]*sys_parts.shape[0]
sr_freq = list(sys_parts["sr_freq"])[0]
sys_parts["ec_freq"]=sys_parts["dummy_sysid"].apply(get_exam_counts)
ec_freq = list(sys_parts["ec_freq"])[0]
parts = set(sys_parts["dummy_part_number"])
counts=sr["dummy_sysid"].groupby(sr.dummy_part_number).count().to_frame().reset_index()
top_replaced=set(counts.head(5))
to_predict = parts.union(top_replaced)
new_row={}
for part in to_predict:
if part in parts:
sys_parts["avg_time"]=sys_parts["dummy_part_number"].apply(avgTimeBetweenServices)
sys_parts["Last_service"]=(pd.Timestamp.now().normalize()-sys_parts['sr_close_date'])/ np.timedelta64(1, 'Y')
else:
new_row={'dummy_sysid':sysid,'dummy_part_number':part,' ageagainstinstallation':age,"avg_time": 0,"sr_freq":sr_freq,"ec_freq":ec_freq,"Last_service":last_sr}
sys_parts.append(new_row,ignore_index=True)
sys_parts = sys_parts.loc[:,["dummy_sysid","dummy_part_number","ageagainstinstallation","avg_time","sr_freq","ec_freq","Last_service"]]
predictions = predict_model(lgbm, data = sys_parts)
predictions["Label"] = pd.qcut(predictions["Label"],q=3,labels=["Red","Yellow","Green"])
red_yellow = predictions[(predictions["Label"]=='Red')|(predictions["Label"]=='Yellow')]
return toSeries(red_yellow)