FarhadPiri-3-classification1

Farhad Piri¶

Crucial Data preprocessings before any algorithms¶

In [1]:

#!pip install zipcodes

In [2]:

#!pip install basemap

In [3]:

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
import zipcodes as zcode
from sklearn.model_selection import train_test_split , GridSearchCV
from sklearn import metrics
from sklearn.linear_model import LogisticRegression, LinearRegression
from sklearn.metrics import classification_report, confusion_matrix
from sklearn.neighbors import KNeighborsClassifier
from sklearn.naive_bayes import ComplementNB
from warnings import filterwarnings
filterwarnings("ignore")

In [4]:

Data = pd.read_csv('Bank_Personal_Loan_Modelling.csv')
df = pd.DataFrame(Data)
df

Out[4]:

	ID	Age	Experience	Income	ZIP Code	Family	CCAvg	Education	Mortgage	Personal Loan	Securities Account	CD Account	Online	CreditCard
0	1	25	1	49	91107	4	1/60	1	0	0	1	0	0	0
1	2	45	19	34	90089	3	1/50	1	0	0	1	0	0	0
2	3	39	15	11	94720	1	1/00	1	0	0	0	0	0	0
3	4	35	9	100	94112	1	2/70	2	0	0	0	0	0	0
4	5	35	8	45	91330	4	1/00	2	0	0	0	0	0	1
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
4995	4996	29	3	40	92697	1	1/90	3	0	0	0	0	1	0
4996	4997	30	4	15	92037	4	0/40	1	85	0	0	0	1	0
4997	4998	63	39	24	93023	2	0/30	3	0	0	0	0	0	0
4998	4999	65	40	49	90034	3	0/50	2	0	0	0	0	1	0
4999	5000	28	4	83	92612	3	0/80	1	0	0	0	0	1	1

5000 rows × 14 columns

In [5]:

df.describe(include='all')

Out[5]:

	ID	Age	Experience	Income	ZIP Code	Family	CCAvg	Education	Mortgage	Personal Loan	Securities Account	CD Account	Online	CreditCard
count	5000.000000	5000.000000	5000.000000	5000.000000	5000.000000	5000.000000	5000	5000.000000	5000.000000	5000.000000	5000.000000	5000.00000	5000.000000	5000.000000
unique	NaN	NaN	NaN	NaN	NaN	NaN	108	NaN	NaN	NaN	NaN	NaN	NaN	NaN
top	NaN	NaN	NaN	NaN	NaN	NaN	0/30	NaN	NaN	NaN	NaN	NaN	NaN	NaN
freq	NaN	NaN	NaN	NaN	NaN	NaN	241	NaN	NaN	NaN	NaN	NaN	NaN	NaN
mean	2500.500000	45.338400	20.104600	73.774200	93152.503000	2.396400	NaN	1.881000	56.498800	0.096000	0.104400	0.06040	0.596800	0.294000
std	1443.520003	11.463166	11.467954	46.033729	2121.852197	1.147663	NaN	0.839869	101.713802	0.294621	0.305809	0.23825	0.490589	0.455637
min	1.000000	23.000000	-3.000000	8.000000	9307.000000	1.000000	NaN	1.000000	0.000000	0.000000	0.000000	0.00000	0.000000	0.000000
25%	1250.750000	35.000000	10.000000	39.000000	91911.000000	1.000000	NaN	1.000000	0.000000	0.000000	0.000000	0.00000	0.000000	0.000000
50%	2500.500000	45.000000	20.000000	64.000000	93437.000000	2.000000	NaN	2.000000	0.000000	0.000000	0.000000	0.00000	1.000000	0.000000
75%	3750.250000	55.000000	30.000000	98.000000	94608.000000	3.000000	NaN	3.000000	101.000000	0.000000	0.000000	0.00000	1.000000	1.000000
max	5000.000000	67.000000	43.000000	224.000000	96651.000000	4.000000	NaN	3.000000	635.000000	1.000000	1.000000	1.00000	1.000000	1.000000

In [6]:

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5000 entries, 0 to 4999
Data columns (total 14 columns):
 #   Column              Non-Null Count  Dtype 
---  ------              --------------  ----- 
 0   ID                  5000 non-null   int64 
 1   Age                 5000 non-null   int64 
 2   Experience          5000 non-null   int64 
 3   Income              5000 non-null   int64 
 4   ZIP Code            5000 non-null   int64 
 5   Family              5000 non-null   int64 
 6   CCAvg               5000 non-null   object
 7   Education           5000 non-null   int64 
 8   Mortgage            5000 non-null   int64 
 9   Personal Loan       5000 non-null   int64 
 10  Securities Account  5000 non-null   int64 
 11  CD Account          5000 non-null   int64 
 12  Online              5000 non-null   int64 
 13  CreditCard          5000 non-null   int64 
dtypes: int64(13), object(1)
memory usage: 547.0+ KB

In [7]:

# Substitute "/" with "." and convert the result to a float value.
df["CCAvg"] = df["CCAvg"].replace("/", ".",regex=True).astype("float")

In [8]:

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 5000 entries, 0 to 4999
Data columns (total 14 columns):
 #   Column              Non-Null Count  Dtype  
---  ------              --------------  -----  
 0   ID                  5000 non-null   int64  
 1   Age                 5000 non-null   int64  
 2   Experience          5000 non-null   int64  
 3   Income              5000 non-null   int64  
 4   ZIP Code            5000 non-null   int64  
 5   Family              5000 non-null   int64  
 6   CCAvg               5000 non-null   float64
 7   Education           5000 non-null   int64  
 8   Mortgage            5000 non-null   int64  
 9   Personal Loan       5000 non-null   int64  
 10  Securities Account  5000 non-null   int64  
 11  CD Account          5000 non-null   int64  
 12  Online              5000 non-null   int64  
 13  CreditCard          5000 non-null   int64  
dtypes: float64(1), int64(13)
memory usage: 547.0 KB

In [9]:

#To avoid negative values in Experience, we will apply the absolute value function.
df['Experience'] = abs(df['Experience'])

In [10]:

df.describe(include="all")

Out[10]:

	ID	Age	Experience	Income	ZIP Code	Family	CCAvg	Education	Mortgage	Personal Loan	Securities Account	CD Account	Online	CreditCard
count	5000.000000	5000.000000	5000.000000	5000.000000	5000.000000	5000.000000	5000.000000	5000.000000	5000.000000	5000.000000	5000.000000	5000.00000	5000.000000	5000.000000
mean	2500.500000	45.338400	20.134600	73.774200	93152.503000	2.396400	1.937938	1.881000	56.498800	0.096000	0.104400	0.06040	0.596800	0.294000
std	1443.520003	11.463166	11.415189	46.033729	2121.852197	1.147663	1.747659	0.839869	101.713802	0.294621	0.305809	0.23825	0.490589	0.455637
min	1.000000	23.000000	0.000000	8.000000	9307.000000	1.000000	0.000000	1.000000	0.000000	0.000000	0.000000	0.00000	0.000000	0.000000
25%	1250.750000	35.000000	10.000000	39.000000	91911.000000	1.000000	0.700000	1.000000	0.000000	0.000000	0.000000	0.00000	0.000000	0.000000
50%	2500.500000	45.000000	20.000000	64.000000	93437.000000	2.000000	1.500000	2.000000	0.000000	0.000000	0.000000	0.00000	1.000000	0.000000
75%	3750.250000	55.000000	30.000000	98.000000	94608.000000	3.000000	2.500000	3.000000	101.000000	0.000000	0.000000	0.00000	1.000000	1.000000
max	5000.000000	67.000000	43.000000	224.000000	96651.000000	4.000000	10.000000	3.000000	635.000000	1.000000	1.000000	1.00000	1.000000	1.000000

In [11]:

# Removing the ID column (as it's unnecessary) and rearranging the target column to be the last.
df_raw = df.drop(columns=["Personal Loan","ID"])
df_raw["Personal Loan"] = df["Personal Loan"]
df_raw.head()

Out[11]:

	Age	Experience	Income	ZIP Code	Family	CCAvg	Education	Securities Account	CreditCard
0	25	1	49	91107	4	1.6	1	1	0
1	45	19	34	90089	3	1.5	1	1	0
2	39	15	11	94720	1	1.0	1	0	0
3	35	9	100	94112	1	2.7	2	0	0
4	35	8	45	91330	4	1.0	2	0	1

In [12]:

# Convert CCAvg to annual
df = df_raw
df['CCAvg'] = df['CCAvg'] * 12
df['CCAvg'] 

Out[12]:

0       19.2
1       18.0
2       12.0
3       32.4
4       12.0
        ... 
4995    22.8
4996     4.8
4997     3.6
4998     6.0
4999     9.6
Name: CCAvg, Length: 5000, dtype: float64

In [13]:

df['ZIP Code'].nunique() 

Out[13]:

Check missing values¶

In [15]:

df.isnull().sum()

Out[15]:

Age                   0
Experience            0
Income                0
ZIP Code              0
Family                0
CCAvg                 0
Education             0
Mortgage              0
Securities Account    0
CD Account            0
Online                0
CreditCard            0
Personal Loan         0
dtype: int64

no missing values

Noise¶

Categorical

In [19]:

Categorical = ['Family', 'Education', 'Personal Loan', 'Securities Account', 'CD Account', 'Online', 'CreditCard']
i = 0
while i<8:
    fig = plt.figure(figsize = [20, 4])
    plt.subplot(1, 3, 1)
    sns.countplot(x = Categorical[i], data = df)
    i += 1
    if i==7:
        break
    plt.subplot(1, 3, 2)
    sns.countplot(x = Categorical[i], data = df)
    i += 1
    plt.subplot(1, 3, 3)
    sns.countplot(x = Categorical[i], data = df)

    i+= 1
    plt.show()

Numerical

In [21]:

Numerical = ['Age', 'Experience', 'CCAvg', 'Mortgage', 'Income']
i = 0
while i<6:
    fig = plt.figure(figsize = [20, 4])
    plt.subplot(1, 2, 1)
    sns.boxplot(x = Numerical[i], data = df)
    i += 1
    if i==5:
        break
    plt.subplot(1, 2, 2)    
    sns.boxplot(x = Numerical[i], data=df)
    i +=1
    plt.show()

Data ranges are okay

Zip Codes

In [24]:

plt.scatter(df['ZIP Code'], df['Personal Loan'])
plt.annotate('maybe noise', xy = (9307, 0), xytext  = (13000, 0.25), arrowprops = dict(facecolor = 'black', shrink = 0.05))
plt.show()

In [25]:

Noise_Zip = df[df['ZIP Code']<20000]
Noise_Zip

Out[25]:

	Age	Experience	Income	ZIP Code	Family	CCAvg	Education	Mortgage	Securities Account	CD Account	Online	CreditCard	Personal Loan
384	51	25	21	9307	4	7.2	3	0	0	0	1	1	0

In [26]:

df.drop(index = 384, inplace = True)

In [27]:

list_zipcode = list(df['ZIP Code'])

In [28]:

county = []
lat=[]
long=[]
for x in list_zipcode:
    my_city_county = zcode.matching(str(x))
    if len(my_city_county)==1:
        County = my_city_county[0].get('county')
        Lat = my_city_county[0].get('lat')
        Long = my_city_county[0].get('long')
    else:
        County = "Notfound"
        Lat = np.NaN
        Long = np.NaN
    county.insert(0, County)
    lat.insert(0, Lat)
    long.insert(0, Long)
county.reverse()
lat.reverse()
long.reverse()

df['Place'] = county
df['Latitude']=lat
df['Longitude'] = long

In [29]:

print(zcode.matching(str(92612)))
df

[{'zip_code': '92612', 'zip_code_type': 'STANDARD', 'active': True, 'city': 'Irvine', 'acceptable_cities': [], 'unacceptable_cities': [], 'state': 'CA', 'county': 'Orange County', 'timezone': 'America/Los_Angeles', 'area_codes': ['714', '949'], 'world_region': 'NA', 'country': 'US', 'lat': '33.6615', 'long': '-117.8217'}]

Out[29]:

	Age	Experience	Income	ZIP Code	Family	CCAvg	Education	Mortgage	Securities Account	CD Account	Online	CreditCard	Personal Loan	Place	Latitude	Longitude
0	25	1	49	91107	4	19.2	1	0	1	0	0	0	0	Los Angeles County	34.1620	-118.0894
1	45	19	34	90089	3	18.0	1	0	1	0	0	0	0	Los Angeles County	34.0218	-118.2883
2	39	15	11	94720	1	12.0	1	0	0	0	0	0	0	Alameda County	37.8746	-122.2547
3	35	9	100	94112	1	32.4	2	0	0	0	0	0	0	San Francisco County	37.7217	-122.4446
4	35	8	45	91330	4	12.0	2	0	0	0	0	1	0	Los Angeles County	34.2429	-118.5273
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
4995	29	3	40	92697	1	22.8	3	0	0	0	1	0	0	Orange County	33.6473	-117.8409
4996	30	4	15	92037	4	4.8	1	85	0	0	1	0	0	San Diego County	32.8668	-117.2482
4997	63	39	24	93023	2	3.6	3	0	0	0	0	0	0	Ventura County	34.5210	-119.2477
4998	65	40	49	90034	3	6.0	2	0	0	0	1	0	0	Los Angeles County	34.0293	-118.3994
4999	28	4	83	92612	3	9.6	1	0	0	0	1	1	0	Orange County	33.6615	-117.8217

4999 rows × 16 columns

926 is Irvine, California, columns are added as expected

In [31]:

df = df.astype({'Latitude':float, 'Longitude':float})

In [32]:

df.describe(include='all')

Out[32]:

	Age	Experience	Income	ZIP Code	Family	CCAvg	Education	Mortgage	Securities Account	CD Account	Online	CreditCard	Personal Loan	Place	Latitude	Longitude
count	4999.000000	4999.000000	4999.000000	4999.000000	4999.000000	4999.000000	4999.000000	4999.000000	4999.000000	4999.000000	4999.000000	4999.000000	4999.000000	4999	4966.000000	4966.000000
unique	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	39	NaN	NaN
top	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	Los Angeles County	NaN	NaN
freq	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	1095	NaN	NaN
mean	45.337267	20.133627	73.784757	93169.275455	2.396079	23.258468	1.880776	56.510102	0.104421	0.060412	0.596719	0.293859	0.096019	NaN	35.792110	-120.037847
std	11.464033	11.416124	46.032281	1759.630610	1.147554	20.972776	0.839804	101.720837	0.305836	0.238273	0.490605	0.455574	0.294647	NaN	2.098821	2.092821
min	23.000000	0.000000	8.000000	90005.000000	1.000000	0.000000	1.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	NaN	32.554700	-124.099400
25%	35.000000	10.000000	39.000000	91911.000000	1.000000	8.400000	1.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	NaN	34.003300	-122.148500
50%	45.000000	20.000000	64.000000	93437.000000	2.000000	18.000000	2.000000	0.000000	0.000000	0.000000	1.000000	0.000000	0.000000	NaN	35.373600	-120.042400
75%	55.000000	30.000000	98.000000	94608.000000	3.000000	30.000000	3.000000	101.000000	0.000000	0.000000	1.000000	1.000000	0.000000	NaN	37.748700	-118.089400
max	67.000000	43.000000	224.000000	96651.000000	4.000000	120.000000	3.000000	635.000000	1.000000	1.000000	1.000000	1.000000	1.000000	NaN	41.758900	-115.630500

Lat = 4966 values¶

Long = 4966 Values¶

the difference says that there are 33 missing values¶

In [34]:

df[df['Place'] == 'NotFound']['ZIP Code'].value_counts()

Out[34]:

Series([], Name: count, dtype: int64)

In [35]:

df['Place'].unique()

Out[35]:

array(['Los Angeles County', 'Alameda County', 'San Francisco County',
       'San Diego County', 'Monterey County', 'Ventura County',
       'Santa Barbara County', 'Marin County', 'Santa Clara County',
       'Santa Cruz County', 'San Mateo County', 'Humboldt County',
       'Contra Costa County', 'Orange County', 'Sacramento County',
       'Yolo County', 'Placer County', 'San Bernardino County',
       'San Luis Obispo County', 'Riverside County', 'Kern County',
       'Notfound', 'Fresno County', 'Sonoma County', 'El Dorado County',
       'San Benito County', 'Butte County', 'Solano County',
       'Mendocino County', 'San Joaquin County', 'Imperial County',
       'Siskiyou County', 'Merced County', 'Trinity County',
       'Stanislaus County', 'Shasta County', 'Tuolumne County',
       'Napa County', 'Lake County'], dtype=object)

In [36]:

def highlight_cols(s):
    color = 'red'
    return 'background-color: %s' % color

5-digit United States postal code #92717, Irvine, CA¶

ZIP : 92717¶

Place Name: Irvine¶

State: California¶

State Abbreviation: CA¶

Coordinates: 33 38 46.32 N, 117 50 23.28 W¶

Latitude: 33.6462 Longitude: -117.8398¶

In [38]:

Z_92717 = list(df[df['ZIP Code']==92717].index)
for x in Z_92717:
    df['Place'].loc[x] = 'Irvine'
    df['Latitude'].loc[x] = 33.6462
    df['Longitude'].loc[x] = -117.8398
df[df['ZIP Code'] == 92717].style.applymap(highlight_cols, subset = pd.IndexSlice[:, ['Personal Loan']])

Out[38]:

	Age	Experience	Income	ZIP Code	Family	CCAvg	Education	Mortgage	Securities Account	Online	CreditCard	Personal Loan	Place	Latitude	Longitude
106	43	17	69	92717	4	34.800000	1	0	0	1	1	0	Irvine	33.646200	-117.839800
172	38	13	171	92717	2	93.600000	1	0	0	1	0	0	Irvine	33.646200	-117.839800
184	52	26	63	92717	2	18.000000	2	0	1	1	0	0	Irvine	33.646200	-117.839800
321	44	20	101	92717	3	52.800000	2	82	0	0	0	1	Irvine	33.646200	-117.839800
366	50	24	35	92717	1	3.600000	3	0	0	1	0	0	Irvine	33.646200	-117.839800
476	60	34	53	92717	1	9.600000	2	0	1	0	1	0	Irvine	33.646200	-117.839800
695	29	4	115	92717	1	22.800000	1	0	0	0	0	0	Irvine	33.646200	-117.839800
721	49	24	39	92717	1	16.800000	3	0	0	1	0	0	Irvine	33.646200	-117.839800
1099	30	6	52	92717	3	8.400000	2	0	0	1	0	0	Irvine	33.646200	-117.839800
1189	42	17	115	92717	2	4.800000	1	0	0	1	0	0	Irvine	33.646200	-117.839800
1483	58	32	63	92717	1	19.200000	1	0	1	1	0	0	Irvine	33.646200	-117.839800
1752	33	8	155	92717	1	88.800000	3	0	0	0	0	1	Irvine	33.646200	-117.839800
1844	65	40	21	92717	3	1.200000	3	0	0	0	1	0	Irvine	33.646200	-117.839800
2049	43	18	94	92717	4	13.200000	2	0	0	1	0	0	Irvine	33.646200	-117.839800
2211	39	14	31	92717	2	16.800000	2	94	0	1	1	0	Irvine	33.646200	-117.839800
2428	39	12	108	92717	4	44.040000	2	301	0	0	1	1	Irvine	33.646200	-117.839800
2486	61	36	130	92717	1	15.600000	1	257	0	0	0	0	Irvine	33.646200	-117.839800
2957	61	36	53	92717	3	6.000000	2	0	0	1	0	0	Irvine	33.646200	-117.839800
4090	42	18	49	92717	3	25.200000	3	0	1	1	0	0	Irvine	33.646200	-117.839800
4276	50	24	155	92717	1	87.600000	1	0	0	1	1	0	Irvine	33.646200	-117.839800
4321	27	0	34	92717	1	24.000000	2	112	0	0	1	0	Irvine	33.646200	-117.839800
4384	45	20	61	92717	3	32.400000	2	0	0	0	1	0	Irvine	33.646200	-117.839800

In [39]:

Z_96651 = list(df[df['ZIP Code']==96651].index)
for x in Z_96651:
    df['Place'].loc[x] = 'Rudno nad Hronom'
    df['Latitude'].loc[x] = 48.4242
    df['Longitude'].loc[x] = 18.7071
df[df['ZIP Code'] == 96651].style.applymap(highlight_cols, subset = pd.IndexSlice[:, ['Personal Loan']])

Out[39]:

	Age	Experience	Income	ZIP Code	Family	CCAvg	Education	Mortgage	Online	CreditCard	Place	Latitude	Longitude
630	32	7	35	96651	3	15.600000	1	108	0	1	Rudno nad Hronom	48.424200	18.707100
672	51	27	23	96651	1	2.400000	1	0	1	0	Rudno nad Hronom	48.424200	18.707100
1426	37	11	60	96651	3	6.000000	3	0	1	0	Rudno nad Hronom	48.424200	18.707100
1653	26	1	24	96651	2	10.800000	3	123	0	1	Rudno nad Hronom	48.424200	18.707100
2731	29	5	28	96651	1	2.400000	3	0	1	0	Rudno nad Hronom	48.424200	18.707100
3525	59	34	13	96651	4	10.800000	2	0	0	0	Rudno nad Hronom	48.424200	18.707100

In [40]:

df_for_mapping = df[df['Place']!= 'Rudno nad Hronom']

In [41]:

#img = plt.imread('map.jpg')

target_0 = df_for_mapping[df_for_mapping['Personal Loan']==0]
target_1 = df_for_mapping[df_for_mapping['Personal Loan']==1]

fig, axs = plt.subplots(figsize = (15, 10))
#axs.imshow(img, extent = [-126, -114, 32, 43], alpha = 0.7)

axs.scatter(target_0['Longitude'], target_0['Latitude'], c= 'cyan', alpha = 0.5, label = 'Loan 0')
axs.scatter(target_1['Longitude'], target_1['Latitude'], c= 'r', alpha = 0.5, label = 'Loan 1')
plt.title('Longitude & Latitude', fontsize = 30, backgroundcolor = 'black', c='white')
plt.xlabel('Longitude', fontsize = 20)
plt.xlabel('Latitude', fontsize = 20)
plt.grid()
plt.legend(loc = 'best')

Out[41]:

<matplotlib.legend.Legend at 0x7f88ecc24a90>

In [42]:

plt.figure(figsize = (15, 10))
sns.countplot(x = 'Place', data = df)
plt.title('Location', fontsize = 30, backgroundcolor  = 'black', c = 'white')
plt.grid()
plt.xticks(rotation = 90, fontsize = 15)
plt.xlabel('place', fontsize = 20)
plt.ylabel('count', fontsize = 20)
plt.show()

In [43]:

plt.figure(figsize = (15, 10))
sns.countplot(y = 'Place', data = df)
plt.title('Location', fontsize = 30, backgroundcolor  = 'black', c = 'white')
plt.grid()
plt.xticks(rotation = 90, fontsize = 15)
plt.xlabel('place', fontsize = 20)
plt.ylabel('count', fontsize = 20)
plt.show()

In [44]:

fig, ax = plt.subplots(figsize = (15, 10))
sns.heatmap(df.corr(numeric_only=True), cmap = 'RdBu_r', cbar = True, annot = True, linewidths = 0.5, ax = ax)
plt.show()

In [45]:

df.corr(numeric_only=True)

Out[45]:

	Age	Experience	Income	ZIP Code	Family	CCAvg	Education	Mortgage	Securities Account	CD Account	Online	CreditCard	Personal Loan	Latitude	Longitude
Age	1.000000	0.993991	-0.055164	-0.030526	-0.046566	-0.051941	0.041211	-0.012484	-0.000403	0.008068	0.013622	0.007530	-0.007694	-0.027757	-0.009929
Experience	0.993991	1.000000	-0.046785	-0.030833	-0.051981	-0.049676	0.013808	-0.011050	-0.000960	0.009757	0.013982	0.008721	-0.008276	-0.028234	-0.008615
Income	-0.055164	-0.046785	1.000000	-0.030725	-0.157232	0.645931	-0.187277	0.206712	-0.002695	0.169703	0.014397	-0.002030	0.502459	-0.031680	-0.023029
ZIP Code	-0.030526	-0.030833	-0.030725	1.000000	0.027532	-0.012197	-0.008254	0.003608	0.002418	0.021669	0.028328	0.024055	-0.002977	0.856855	-0.261323
Family	-0.046566	-0.051981	-0.157232	0.027532	1.000000	-0.109088	0.064581	-0.020294	0.020094	0.014184	0.010127	0.011160	0.061471	0.026448	-0.012197
CCAvg	-0.051941	-0.049676	0.645931	-0.012197	-0.109088	1.000000	-0.135952	0.109829	0.015035	0.136504	-0.003486	-0.006454	0.366864	-0.019101	-0.020693
Education	0.041211	0.013808	-0.187277	-0.008254	0.064581	-0.135952	1.000000	-0.033186	-0.010723	0.014004	-0.015227	-0.011432	0.136834	-0.013418	0.015991
Mortgage	-0.012484	-0.011050	0.206712	0.003608	-0.020294	0.109829	-0.033186	1.000000	-0.005449	0.089286	-0.005904	-0.007061	0.142065	-0.003181	-0.004485
Securities Account	-0.000403	-0.000960	-0.002695	0.002418	0.020094	0.015035	-0.010723	-0.005449	1.000000	0.317023	0.012685	-0.014926	0.021932	-0.004530	-0.012094
CD Account	0.008068	0.009757	0.169703	0.021669	0.014184	0.136504	0.014004	0.089286	0.317023	1.000000	0.175935	0.278792	0.316344	0.023647	-0.020580
Online	0.013622	0.013982	0.014397	0.028328	0.010127	-0.003486	-0.015227	-0.005904	0.012685	0.175935	1.000000	0.003956	0.006332	0.028896	-0.014014
CreditCard	0.007530	0.008721	-0.002030	0.024055	0.011160	-0.006454	-0.011432	-0.007061	-0.014926	0.278792	0.003956	1.000000	0.002903	0.015377	-0.005542
Personal Loan	-0.007694	-0.008276	0.502459	-0.002977	0.061471	0.366864	0.136834	0.142065	0.021932	0.316344	0.006332	0.002903	1.000000	-0.008494	-0.010898
Latitude	-0.027757	-0.028234	-0.031680	0.856855	0.026448	-0.019101	-0.013418	-0.003181	-0.004530	0.023647	0.028896	0.015377	-0.008494	1.000000	-0.183501
Longitude	-0.009929	-0.008615	-0.023029	-0.261323	-0.012197	-0.020693	0.015991	-0.004485	-0.012094	-0.020580	-0.014014	-0.005542	-0.010898	-0.183501	1.000000

In [46]:

plt.figure(figsize = (15, 10))
sns.kdeplot(data = df, x = 'Income', y = 'Age', hue = 'Personal Loan', shade = True, fill = True, common_norm =False,
            palette= 'crest', alpha = 0.5, linewidth = 0)
plt.title('Income-Age', fontsize = 20)
plt.xlabel('Income', fontsize = 20)
plt.ylabel ('Age', fontsize = 20)
plt.xticks(fontsize = 20)
plt.yticks(fontsize=20)
plt.show()

In [47]:

plt.figure(figsize = (15, 10))
sns.kdeplot(data = df, x = 'Age', y = 'CCAvg', hue = 'Personal Loan', shade = True, fill = True, common_norm =False,
            palette= 'crest', alpha = 0.5, linewidth = 0)
plt.title('Age-CCAvg', fontsize = 20)
plt.xlabel('Age', fontsize = 20)
plt.ylabel ('CCAvg', fontsize = 20)
plt.xticks(fontsize = 20)
plt.yticks(fontsize=20)
plt.show()

In [100]:

X[X['Latitude'].isnull()]

Out[100]:

	Age	Experience	Income	ZIP Code	Family	CCAvg	Education	Securities Account	Online	CreditCard	Latitude	Longitude
468	34	10	21	92634	1	6.0	3	0	1	0	NaN	NaN
780	32	7	42	92634	4	9.6	1	0	1	1	NaN	NaN
2218	38	13	9	92634	2	3.6	2	0	0	0	NaN	NaN
3887	24	2	118	92634	2	86.4	1	1	1	0	NaN	NaN
4392	52	27	81	92634	4	45.6	2	0	0	0	NaN	NaN

In [134]:

df2 = df.dropna()
df2.reset_index(drop= True, inplace = True)

X = df2.drop(columns = ['Personal Loan', 'Place'], axis =1)


y =df2['Personal Loan'].values.reshape(-1, 1)
X_train, X_test, y_train, y_test= train_test_split(X1, y, test_size = 0.2, random_state = 0)

LogReg¶

In [148]:

logreg = LogisticRegression(solver = 'liblinear', C=50)
logreg.fit(X_train, y_train.ravel())
y_pred = logreg.predict(X_test)
print('f1 score is: ', metrics.f1_score(y_test, y_pred))

f1 score is:  0.37762237762237766

In [150]:

def Replace(dataframe):
    dataframe = df2.copy()
    dataframe['Education'].replace([1,2,3], [4, 5, 6], inplace= True)
    dataframe['Securities Account'].replace([1,2], [3, 4], inplace= True)
    dataframe['CD Account'].replace([1,2], [3, 4], inplace= True)
    dataframe['Online'].replace([1, 2], [3, 4], inplace= True)    
    return(dataframe)

In [285]:

def Logreg(X, y, Testsize):
    df_evaluation=pd.DataFrame()
    for x in Testsize:
        X_train, X_test, y_train, y_test= train_test_split(X, y, test_size = x, random_state = 0)
        logreg = LogisticRegression(solver = 'liblinear', class_weight= 'balanced')
        logreg.fit(X_train, y_train)
        y_pred = logreg.predict(X_test)
        dict = {'Test_size': x, 'acc': metrics.accuracy_score(y_test, y_pred), "score":logreg.score(X, y)}
        dict_df = pd.DataFrame([dict])

        df_evaluation = pd.concat([df_evaluation, dict_df], ignore_index = True)
    return (X_train, X_test, y_train, y_test, y_pred, df_evaluation)
def highlight_max(s):
    is_max = s == s.max()
    return ['background-color: yellow' if v else '' for v in is_max]
        

In [287]:

df_r = Replace(df2)
X = df_r.drop(columns=['Personal Loan', 'Place'])
y = df_r['Personal Loan']

In [289]:

X_train, X_test, y_train, y_test, y_pred, df_evaluation= Logreg(X, y, [.1, .15, .2, .25, .3])
df_evaluation.style.apply(highlight_max)

Out[289]:

	Test_size	acc	score
0	0.100000	0.892000	0.893472
1	0.150000	0.893333	0.894473
2	0.200000	0.894895	0.892070
3	0.250000	0.884708	0.889668
4	0.300000	0.887258	0.890669

In [214]:

fpr, tpr, _ = metrics.roc_curve(y_test, y_pred)
plt.plot(fpr, tpr, label = "data1")
plt.legend(loc = 4)
plt.show()

In [220]:

y_pred_proba = logreg.predict_proba(X_test)[::, 1]
fpr, tpr, _ = metrics.roc_curve(y_test, y_pred_proba)
plt.plot(fpr, tpr, label = "data1")
plt.legend(loc = 'best')
plt.show()

In [226]:

confusion_matrix(y_test, y_pred)

Out[226]:

array([[1210,  156],
       [  13,  120]])

In [228]:

solvers = ['newton-cg', 'liblinear', 'lbfgs', 'sag', 'saga']

newton-cg¶

In [301]:

def Logreg(X, y, Testsize):
    df_evaluation=pd.DataFrame()
    for x in Testsize:
        X_train, X_test, y_train, y_test= train_test_split(X, y, test_size = x, random_state = 0)
        logreg = LogisticRegression(solver = 'newton-cg', class_weight= 'balanced')
        logreg.fit(X_train, y_train)
        y_pred = logreg.predict(X_test)
        dict = {'Test_size': x, 'acc': metrics.accuracy_score(y_test, y_pred), "score":logreg.score(X, y)}
        dict_df = pd.DataFrame([dict])

        df_evaluation = pd.concat([df_evaluation, dict_df], ignore_index = True)
    return (X_train, X_test, y_train, y_test, y_pred, df_evaluation)
def highlight_max(s):
    is_max = s == s.max()
    return ['background-color: yellow' if v else '' for v in is_max]

In [303]:

X_train, X_test, y_train, y_test, y_pred, df_evaluation= Logreg(X, y, [.1, .15, .2, .25, .3])
df_evaluation.style.apply(highlight_max)

Out[303]:

	Test_size	acc	score
0	0.100000	0.892000	0.895675
1	0.150000	0.897333	0.896476
2	0.200000	0.896897	0.895675
3	0.250000	0.886309	0.891670
4	0.300000	0.891261	0.893272

lbfgs¶

In [305]:

def Logreg(X, y, Testsize):
    df_evaluation=pd.DataFrame()
    for x in Testsize:
        X_train, X_test, y_train, y_test= train_test_split(X, y, test_size = x, random_state = 0)
        logreg = LogisticRegression(solver = 'lbfgs', class_weight= 'balanced')
        logreg.fit(X_train, y_train)
        y_pred = logreg.predict(X_test)
        dict = {'Test_size': x, 'acc': metrics.accuracy_score(y_test, y_pred), "score":logreg.score(X, y)}
        dict_df = pd.DataFrame([dict])

        df_evaluation = pd.concat([df_evaluation, dict_df], ignore_index = True)
    return (X_train, X_test, y_train, y_test, y_pred, df_evaluation)
def highlight_max(s):
    is_max = s == s.max()
    return ['background-color: yellow' if v else '' for v in is_max]

In [307]:

X_train, X_test, y_train, y_test, y_pred, df_evaluation= Logreg(X, y, [.1, .15, .2, .25, .3])
df_evaluation.style.apply(highlight_max)

Out[307]:

	Test_size	acc	score
0	0.100000	0.850000	0.856628
1	0.150000	0.890667	0.890268
2	0.200000	0.879880	0.879856
3	0.250000	0.881505	0.887665
4	0.300000	0.859239	0.859632

sag¶

In [313]:

def Logreg(X, y,Testsize):
    df_evaluation=pd.DataFrame()
    for x in Testsize:
        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=x, random_state =0)
        logreg = LogisticRegression(solver = 'sag', class_weight= 'balanced')
        logreg.fit(X_train, y_train)
        y_pred = logreg.predict(X_test)
        dict = {'Test_Size': x, 'acc': metrics.accuracy_score(y_test, y_pred), 'score':logreg.score(X, y)}
        dict_df = pd.DataFrame([dict])
        df_evaluation = pd.concat([df_evaluation, dict_df], ignore_index =True)
    return (X_train, X_test, y_train, y_test, y_pred, df_evaluation)
def highlight_max(s):
    is_max = s == s.max()
    return ['background-color: yellow' if v else '' for v in is_max]

In [315]:

X_train, X_test, y_train, y_test, y_pred, df_evaluation= Logreg(X, y, [.1, .15, .2, .25, .3])
df_evaluation.style.apply(highlight_max)

Out[315]:

	Test_Size	acc	score
0	0.100000	0.912000	0.903885
1	0.150000	0.421333	0.417901
2	0.200000	0.907908	0.903484
3	0.250000	0.800641	0.810773
4	0.300000	0.911274	0.903885

saga¶

In [317]:

def Logreg(X, y,Testsize):
    df_evaluation=pd.DataFrame()
    for x in Testsize:
        X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=x, random_state =0)
        logreg = LogisticRegression(solver = 'saga', class_weight= 'balanced')
        logreg.fit(X_train, y_train)
        y_pred = logreg.predict(X_test)
        dict = {'Test_Size': x, 'acc': metrics.accuracy_score(y_test, y_pred), 'score':logreg.score(X, y)}
        dict_df = pd.DataFrame([dict])
        df_evaluation = pd.concat([df_evaluation, dict_df], ignore_index =True)
    return (X_train, X_test, y_train, y_test, y_pred, df_evaluation)
def highlight_max(s):
    is_max = s == s.max()
    return ['background-color: yellow' if v else '' for v in is_max]

In [319]:

X_train, X_test, y_train, y_test, y_pred, df_evaluation= Logreg(X, y, [.1, .15, .2, .25, .3])
df_evaluation.style.apply(highlight_max)

Out[319]:

	Test_Size	acc	score
0	0.100000	0.742000	0.751902
1	0.150000	0.864000	0.853624
2	0.200000	0.832833	0.828594
3	0.250000	0.909528	0.904686
4	0.300000	0.903269	0.893072

GridSearch¶

In [322]:

from sklearn.model_selection import KFold
logreg2 = LogisticRegression(solver = 'newton-cg', max_iter = 5000, n_jobs = -1)
params = {'penalty':['12', 'none'], 'C': [0.1, 0.2, 0.5, 1, 2, 5, 10, 20, 50, 100, 200],
        'class_weight': [None, 'balanced'], 'warm_start':[False, True]}
kfold = KFold(n_splits = 5, shuffle =True, random_state =  1)
gridsearch = GridSearchCV(logreg2, param_grid=params, scoring = 'f1', n_jobs = -1, cv =kfold)

In [ ]:

import warnings

warnings.filterwarnings('ignore')
warnings.simplefilter(action='ignore')

gridsearch.fit(X_train, y_train)

In [341]:

print(gridsearch.best_params_)
print(gridsearch.best_score_)

{'C': 0.1, 'class_weight': None, 'penalty': 'none', 'warm_start': False}
0.7184085514452588

KNN¶

In [397]:

training_acc = []
test_acc = []
neighbors_setting = range(1, 30)
for n_neighbors in neighbors_setting:
    knn_model = KNeighborsClassifier(n_neighbors = n_neighbors)
    knn_model.fit(X_train, y_train.ravel())
    training_acc.append(knn_model.score(X_train, y_train))
    test_acc.append(knn_model.score(X_test, y_test))
plt.figure(figsize = (15, 10))
plt.plot(neighbors_setting, training_acc, label = "Accuracy of the training set")
plt.plot(neighbors_setting, test_acc, label = "Accuracy of the test set")
plt.ylabel('Accuracy')
plt.xlabel('Number of Neighbors')
plt.grid()
plt.legend()

Out[397]:

<matplotlib.legend.Legend at 0x7f88d6b1d310>

In [399]:

print(np.max(training_acc))
print(np.max(test_acc))

1.0
0.914609739826551

In [401]:

knn_model = KNeighborsClassifier(9)
knn_model.fit(X_train, y_train.ravel())
y_pred = knn_model.predict(X_test)
print("Accuracy for knn: ", metrics.accuracy_score(y_test, y_pred))

Accuracy for knn:  0.9092728485657104

In [403]:

confusion_matrix(y_test, y_pred)

Out[403]:

array([[1338,   28],
       [ 108,   25]])

This con_matrix is for LogRes

In [406]:

from sklearn.model_selection import GridSearchCV

parameters = {"n_neighbors": range(1, 30) }
gridsearch2 = GridSearchCV(estimator = knn_model, 
                          param_grid = parameters,
                          scoring = 'f1',
                          cv = 5,
                          verbose = 1,
                          n_jobs = -1)
gridsearch2.fit(X_train, y_train.ravel())
gridsearch2.best_params_

Fitting 5 folds for each of 29 candidates, totalling 145 fits

Out[406]:

{'n_neighbors': 1}

In [407]:

print(gridsearch2.best_params_)
print(gridsearch2.best_score_)

{'n_neighbors': 1}
0.3653787148242307

GridSearch for logres

Naive Bayes¶

Complement

In [411]:

x = df.drop(columns=['Personal Loan', 'Place', 'Longitude', 'Latitude'])
y=df['Personal Loan']
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 0)
clf = ComplementNB()
clf.fit(X_train, y_train.ravel())
y_pred = clf.predict(X_test)
print('Accuracy is', metrics.accuracy_score(y_test, y_pred))

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[411], line 3
      1 x = df.drop(columns=['Personal Loan', 'Place', 'Longitude', 'Latitude'])
      2 y=df['Personal Loan']
----> 3 X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2, random_state = 0)
      4 clf = ComplementNB()
      5 clf.fit(X_train, y_train.ravel())

File /opt/conda/envs/anaconda-panel-2023.05-py310/lib/python3.11/site-packages/sklearn/utils/_param_validation.py:211, in validate_params.<locals>.decorator.<locals>.wrapper(*args, **kwargs)
    205 try:
    206     with config_context(
    207         skip_parameter_validation=(
    208             prefer_skip_nested_validation or global_skip_validation
    209         )
    210     ):
--> 211         return func(*args, **kwargs)
    212 except InvalidParameterError as e:
    213     # When the function is just a wrapper around an estimator, we allow
    214     # the function to delegate validation to the estimator, but we replace
    215     # the name of the estimator by the name of the function in the error
    216     # message to avoid confusion.
    217     msg = re.sub(
    218         r"parameter of \w+ must be",
    219         f"parameter of {func.__qualname__} must be",
    220         str(e),
    221     )

File /opt/conda/envs/anaconda-panel-2023.05-py310/lib/python3.11/site-packages/sklearn/model_selection/_split.py:2614, in train_test_split(test_size, train_size, random_state, shuffle, stratify, *arrays)
   2611 if n_arrays == 0:
   2612     raise ValueError("At least one array required as input")
-> 2614 arrays = indexable(*arrays)
   2616 n_samples = _num_samples(arrays[0])
   2617 n_train, n_test = _validate_shuffle_split(
   2618     n_samples, test_size, train_size, default_test_size=0.25
   2619 )

File /opt/conda/envs/anaconda-panel-2023.05-py310/lib/python3.11/site-packages/sklearn/utils/validation.py:455, in indexable(*iterables)
    436 """Make arrays indexable for cross-validation.
    437 
    438 Checks consistent length, passes through None, and ensures that everything
   (...)
    451     sparse matrix, or dataframe) or `None`.
    452 """
    454 result = [_make_indexable(X) for X in iterables]
--> 455 check_consistent_length(*result)
    456 return result

File /opt/conda/envs/anaconda-panel-2023.05-py310/lib/python3.11/site-packages/sklearn/utils/validation.py:409, in check_consistent_length(*arrays)
    407 uniques = np.unique(lengths)
    408 if len(uniques) > 1:
--> 409     raise ValueError(
    410         "Found input variables with inconsistent numbers of samples: %r"
    411         % [int(l) for l in lengths]
    412     )

ValueError: Found input variables with inconsistent numbers of samples: [4994, 4999]

Naive Bayes alg. is not suitable for this dataset

In [392]:

df['Personal Loan'].value_counts()

Out[392]:

Personal Loan
0    4519
1     480
Name: count, dtype: int64

In [394]:

4519/480

Out[394]:

9.414583333333333

Imbalanced distribution (1:9.41)

How to process data before any ML task