Commit
•
0a4a484
1
Parent(s):
1107b8f
Update data_preparation.py
Browse files- data_preparation.py +83 -20
data_preparation.py
CHANGED
@@ -1,20 +1,83 @@
|
|
1 |
-
import pandas as pd
|
2 |
-
import numpy as np
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
-
|
20 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import pandas as pd
|
2 |
+
import numpy as np
|
3 |
+
def data_imp():
|
4 |
+
feature_descriptions = {
|
5 |
+
"CustID": "Unique identifier for each customer.",
|
6 |
+
"FirstPolYear": "Year when the customer first bought an insurance policy.",
|
7 |
+
"BirthYear": "Birth year of the customer, used to calculate age.",
|
8 |
+
"EducDeg": "Highest educational degree obtained by the customer.",
|
9 |
+
"MonthSal": "Monthly salary of the customer. (Numerical, float64)",
|
10 |
+
"GeoLivArea": "Geographical area where the customer lives.",
|
11 |
+
"Children": "Number of children the customer has.",
|
12 |
+
"CustMonVal": "Total monetary value of the customer to the company.",
|
13 |
+
"ClaimsRate": "Rate at which the customer files insurance claims.",
|
14 |
+
"PremMotor": "Premium amount for motor insurance.",
|
15 |
+
"PremHousehold": "Premium amount for household insurance.",
|
16 |
+
"PremHealth": "Premium amount for health insurance.",
|
17 |
+
"PremLife": "Premium amount for life insurance.",
|
18 |
+
"PremWork": "Premium amount for work insurance."
|
19 |
+
}
|
20 |
+
insurance_defaults = {
|
21 |
+
"FirstPolYear": 1999,
|
22 |
+
"BirthYear": 1980,
|
23 |
+
"MonthSal": 1000,
|
24 |
+
"GeoLivArea": 0, # Options: 0, 1, 2, 3
|
25 |
+
"Children": 0, # Options: 0, 1, 2
|
26 |
+
"CustMonVal": 100,
|
27 |
+
"ClaimsRate": 2.33,
|
28 |
+
"PremMotor": 200,
|
29 |
+
"PremHousehold": 200,
|
30 |
+
"PremHealth": 200,
|
31 |
+
"PremLife": 200,
|
32 |
+
"PremWork": 200
|
33 |
+
}
|
34 |
+
|
35 |
+
# Define default values for banking dataset features
|
36 |
+
banking_defaults = {
|
37 |
+
"BALANCE": 2000,
|
38 |
+
"BALANCE_FREQUENCY": 0.5,
|
39 |
+
"PURCHASES": 500,
|
40 |
+
"ONEOFF_PURCHASES": 0,
|
41 |
+
"INSTALLMENTS_PURCHASES": 0,
|
42 |
+
"CASH_ADVANCE": 200,
|
43 |
+
"PURCHASES_FREQUENCY": 0.1,
|
44 |
+
"ONEOFF_PURCHASES_FREQUENCY": 0.1,
|
45 |
+
"PURCHASES_INSTALLMENTS_FREQUENCY": 0.5,
|
46 |
+
"CASH_ADVANCE_FREQUENCY": 5,
|
47 |
+
"CASH_ADVANCE_TRX": 5,
|
48 |
+
"PURCHASES_TRX": 5,
|
49 |
+
"CREDIT_LIMIT": 10000,
|
50 |
+
"PAYMENTS": 500,
|
51 |
+
"MINIMUM_PAYMENTS": 130,
|
52 |
+
"PRC_FULL_PAYMENT": 0.22,
|
53 |
+
"TENURE": 10
|
54 |
+
}
|
55 |
+
|
56 |
+
# Define default values for retail dataset features
|
57 |
+
retail_defaults = {
|
58 |
+
"Fresh": 6000,
|
59 |
+
"Milk": 9000,
|
60 |
+
"Grocery": 9000,
|
61 |
+
"Frozen": 4000,
|
62 |
+
"Detergents_Paper": 4000,
|
63 |
+
"Delicassen": 2000
|
64 |
+
}
|
65 |
+
return feature_descriptions,insurance_defaults,banking_defaults,retail_defaults
|
66 |
+
|
67 |
+
def preprocess_data(data):
|
68 |
+
if 'CustID' in data.columns:
|
69 |
+
data = data.drop(columns=['CustID'])
|
70 |
+
if 'Channel' in data.columns:
|
71 |
+
data = data.drop(columns=['Channel'])
|
72 |
+
if 'Region' in data.columns:
|
73 |
+
data = data.drop(columns=['Region'])
|
74 |
+
|
75 |
+
|
76 |
+
data = remove_outliers(data)
|
77 |
+
return data
|
78 |
+
|
79 |
+
def remove_outliers(df, threshold=3):
|
80 |
+
df_numeric = df.select_dtypes(include=[float, int])
|
81 |
+
z_scores = np.abs((df_numeric - df_numeric.mean()) / df_numeric.std())
|
82 |
+
df_clean = df[(z_scores < threshold).all(axis=1)]
|
83 |
+
return df_clean
|