7/28/23, 1:10 PM Assignment - Jupyter Notebook

Assignment
You have been provided with a dataset containing information about customers of an e-commerce company. The task is
to build a binary classification model using logistic regression to predict whether a customer will make a purchase or not
based on their demographic and browsing behavior data. The dataset consists of the following features:

email
address
avatar
time on app
time on website
length of membership
yearly amount spent

The target variable is: Purchase (binary: 1 if the customer made a purchase over $450, 0 otherwise)

Instructions: Load the dataset and perform any necessary data preprocessing steps. Split the data into training and
testing sets (e.g., 80% training, 20% testing). Train a logistic regression model using the training data. Evaluate the
model's performance on the testing data using appropriate evaluation metrics (e.g., accuracy, precision, recall, F1-score).

Provide a brief summary of the model's performance and any insights you gather from the results. Note: You can use any
programming language or machine learning libraries of your choice. The aim of this problem is to assess your ability to
quickly understand the problem, preprocess the data, build a logistic regression model, evaluate its performance, and
derive meaningful insights from the results within a limited timeframe.

In [340]:

1 import numpy as np
2 import pandas as pd
3 import matplotlib.pyplot as plt
4 from sklearn.preprocessing import StandardScaler
5 from imblearn.over_sampling import RandomOverSampler
6
7 %matplotlib inline

In [341]:

1 df = pd.read_csv("scpl_folder/data.csv")

localhost:8888/notebooks/Desktop/Personal/code/Practice_ML/Assignment.ipynb# 1/10
7/28/23, 1:10 PM Assignment - Jupyter Notebook

In [342]:

1 df.head(10)

Out[342]:

Time Yearly
Time on Length of
\tEmail Address Avatar on Amount C
Website Membership
App Spent

16338 Scott Corner Suite

0 [email protected] SeaGreen 10.16 37.76 4.78 521.24
727West Alexandra, AR...

672 Jesus Roads Apt.

1 [email protected] 443Thompsonland, WY LightSkyBlue 13.46 37.24 2.94 503.98
69228

38678 Sean Drive Suite

2 [email protected] 293Karentown, IA DarkGray 12.01 36.53 4.71 576.48
78306-...

0128 Sampson Loop

3 [email protected] Suite 943Hoffmanton, SaddleBrown 10.10 38.04 4.24 418.60
MO 02122

5791 Jessica
acampbell@sanchez-
4 CoveMckinneyborough, Wheat 11.45 37.58 2.59 420.74
velasquez.info
OK 64460-7536

88995 Edwards Row

5 [email protected] Suite 456North Jo, DE Sienna 10.74 37.46 3.86 476.19
02062-...

9991 Macdonald
6 [email protected] SquaresVasquezborough, Purple 10.97 36.61 2.87 404.82
WY 73586...

2595 James Creek Apt.

7 [email protected] PaleVioletRed 11.76 37.92 3.53 482.14
571Millerberg, HI 82236

399 Jeremy Skyway

8 [email protected] Suite 377North Keithville, PaleTurquoise 12.19 36.15 3.78 494.55
I...

PSC 2490, Box 2120APO

9 [email protected] Black 12.88 37.44 1.56 419.94
AE 15445-2876

In [343]:

1 df.columns = ['Email', 'Address', 'Avatar', 'Time on App', 'Time on Website',

2 'Length of Membership', 'Yearly Amount Spent', 'Clean_Address_Loc','Clean_Address_County

In [344]:

1 df.loc[df['Yearly Amount Spent']>450,'purchase']=1

2 df.loc[df['Yearly Amount Spent']<=450,'purchase']=0

localhost:8888/notebooks/Desktop/Personal/code/Practice_ML/Assignment.ipynb# 2/10
7/28/23, 1:10 PM Assignment - Jupyter Notebook

In [345]:

1 df.describe()

Out[345]:

Time on App Time on Website Length of Membership Yearly Amount Spent purchase

count 500.000000 500.000000 500.00000 500.000000 500.000000

mean 12.052620 37.060480 3.53336 499.314240 0.730000

std 0.994418 1.010555 0.99926 79.314764 0.444404

min 8.510000 33.910000 0.27000 256.670000 0.000000

25% 11.390000 36.347500 2.93000 445.037500 0.000000

50% 11.980000 37.070000 3.53500 498.890000 1.000000

75% 12.752500 37.720000 4.13000 549.312500 1.000000

max 15.130000 40.010000 6.92000 765.520000 1.000000

In [346]:

1 df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 500 entries, 0 to 499
Data columns (total 10 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 Email 500 non-null object
1 Address 500 non-null object
2 Avatar 500 non-null object
3 Time on App 500 non-null float64
4 Time on Website 500 non-null float64
5 Length of Membership 500 non-null float64
6 Yearly Amount Spent 500 non-null float64
7 Clean_Address_Loc 500 non-null object
8 Clean_Address_County 500 non-null object
9 purchase 500 non-null float64
dtypes: float64(5), object(5)
memory usage: 39.2+ KB

In [347]:

1 df.corr()

Out[347]:

Time on Time on Length of Yearly Amount

purchase
App Website Membership Spent

Time on App 1.000000 0.082285 0.029240 0.499315 0.353636

Time on Website 0.082285 1.000000 -0.047443 -0.002601 0.003681

Length of
0.029240 -0.047443 1.000000 0.809184 0.601839
Membership

Yearly Amount Spent 0.499315 -0.002601 0.809184 1.000000 0.737246

purchase 0.353636 0.003681 0.601839 0.737246 1.000000

localhost:8888/notebooks/Desktop/Personal/code/Practice_ML/Assignment.ipynb# 3/10
7/28/23, 1:10 PM Assignment - Jupyter Notebook

Vectorize the words

In [289]:

1 df = df.drop(['Email','Address','Clean_Address_Loc'], axis=1)
2 df

Out[289]:

Yearly
Time on Time on Length of
Avatar Amount Clean_Address_County purchase
App Website Membership
Spent

0 SeaGreen 10.16 37.76 4.78 521.24 AR 1.0

1 LightSkyBlue 13.46 37.24 2.94 503.98 WY 1.0

2 DarkGray 12.01 36.53 4.71 576.48 IA 1.0

3 SaddleBrown 10.10 38.04 4.24 418.60 MO 0.0

4 Wheat 11.45 37.58 2.59 420.74 OK 0.0

... ... ... ... ... ... ... ...

495 DodgerBlue 12.94 36.73 4.56 544.41 UT 1.0

496 OldLace 11.83 36.84 3.61 502.09 MI 1.0

497 Purple 11.68 38.72 3.59 463.59 MT 1.0

498 Moccasin 12.75 36.71 3.28 548.28 Bo 1.0

499 PeachPuff 12.13 38.19 4.02 597.74 SC 1.0

500 rows × 7 columns

In [290]:

1 from sklearn.model_selection import train_test_split

2 from sklearn.ensemble import RandomForestClassifier
3 from sklearn.metrics import accuracy_score
4 from sklearn.feature_extraction.text import CountVectorizer
5 vectorizer = CountVectorizer(analyzer = "word",
6 tokenizer = None,
7 preprocessor = None,
8 stop_words = None,
9 max_features = 2)

In [291]:

1 def vect(df, col):

2 dfw = vectorizer.fit_transform(df[col])
3 dfw = dfw.toarray()
4 df = np.hstack((df.drop(col, axis =1),np.reshape(dfw,(-1,2))))
5 df= pd.DataFrame(df)
6 print(df)
7 return df

localhost:8888/notebooks/Desktop/Personal/code/Practice_ML/Assignment.ipynb# 4/10
7/28/23, 1:10 PM Assignment - Jupyter Notebook

In [292]:

1 df = vect(df,'Avatar')
2 df.columns =['Time on App', 'Time on Website','Length of Membership', 'Yearly Amount Spent','Cl
3 df = vect(df,'Clean_Address_County')
4 df.columns =['Time on App', 'Time on Website','Length of Membership', 'Yearly Amount Spent','pu
5 df.head()

0 1 2 3 4 5 6 7
0 10.16 37.76 4.78 521.24 AR 1.0 0 0
1 13.46 37.24 2.94 503.98 WY 1.0 0 0
2 12.01 36.53 4.71 576.48 IA 1.0 0 0
3 10.1 38.04 4.24 418.6 MO 0.0 0 0
4 11.45 37.58 2.59 420.74 OK 0.0 0 0
.. ... ... ... ... .. ... .. ..
495 12.94 36.73 4.56 544.41 UT 1.0 0 0
496 11.83 36.84 3.61 502.09 MI 1.0 0 0
497 11.68 38.72 3.59 463.59 MT 1.0 0 0
498 12.75 36.71 3.28 548.28 Bo 1.0 0 0
499 12.13 38.19 4.02 597.74 SC 1.0 0 0

[500 rows x 8 columns]

0 1 2 3 4 5 6 7 8
0 10.16 37.76 4.78 521.24 1.0 0 0 0 0
1 13.46 37.24 2.94 503.98 1.0 0 0 0 0
2 12.01 36.53 4.71 576.48 1.0 0 0 0 0
3 10.1 38.04 4.24 418.6 0.0 0 0 0 0
4 11.45 37.58 2.59 420.74 0.0 0 0 0 0
.. ... ... ... ... ... .. .. .. ..
495 12.94 36.73 4.56 544.41 1.0 0 0 0 0
496 11.83 36.84 3.61 502.09 1.0 0 0 0 0
497 11.68 38.72 3.59 463.59 1.0 0 0 0 0
498 12.75 36.71 3.28 548.28 1.0 0 0 1 0
499 12.13 38.19 4.02 597.74 1.0 0 0 0 1

[500 rows x 9 columns]

Out[292]:

Time on Time on Length of Yearly Amount

purchase avatar1 avatar2 clc1 clc2
App Website Membership Spent

0 10.16 37.76 4.78 521.24 1.0 0 0 0 0

1 13.46 37.24 2.94 503.98 1.0 0 0 0 0

2 12.01 36.53 4.71 576.48 1.0 0 0 0 0

3 10.1 38.04 4.24 418.6 0.0 0 0 0 0

4 11.45 37.58 2.59 420.74 0.0 0 0 0 0

In [294]:

1 train,test = np.split(df.sample(frac=1),[int(0.8*len(df))])

In [295]:

1 train.shape, test.shape

Out[295]:

((400, 9), (100, 9))

localhost:8888/notebooks/Desktop/Personal/code/Practice_ML/Assignment.ipynb# 5/10
7/28/23, 1:10 PM Assignment - Jupyter Notebook

In [296]:

1 train = pd.DataFrame(train)
2 train.columns =df.columns
3 print(train.head())
4
5 test = pd.DataFrame(test)
6 test.columns =df.columns
7 print(test.head())

Time on App Time on Website Length of Membership Yearly Amount Spent \

235 11.08 37.96 4.72 517.17
11 12.6 37.37 3.47 501.93
55 12.36 38.04 3.31 468.91
473 11.47 35.68 1.81 374.27
201 12.52 37.15 2.67 487.38

purchase avatar1 avatar2 clc1 clc2

235 1.0 1 0 0 0
11 1.0 0 0 0 0
55 1.0 0 0 1 0
473 0.0 0 0 0 0
201 1.0 0 0 0 0
Time on App Time on Website Length of Membership Yearly Amount Spent \
97 12.91 36.05 3.49 547.71
496 11.83 36.84 3.61 502.09
57 11.17 35.63 5.46 587.57
334 13.29 38.63 3.87 543.34
95 11.33 35.46 4.54 568.72

purchase avatar1 avatar2 clc1 clc2

97 1.0 0 0 0 0
496 1.0 0 0 0 0
57 1.0 0 0 0 0
334 1.0 0 0 0 0
95 1.0 0 0 0 0

localhost:8888/notebooks/Desktop/Personal/code/Practice_ML/Assignment.ipynb# 6/10
7/28/23, 1:10 PM Assignment - Jupyter Notebook

In [297]:

1 def column_to_move(df):
2 column_to_move = df.pop("purchase")
3 df.insert(8, "purchase", column_to_move)
4 return df
5
6 train = column_to_move(train)
7 test = column_to_move(test)
8
9 train,test

localhost:8888/notebooks/Desktop/Personal/code/Practice_ML/Assignment.ipynb# 7/10
7/28/23, 1:10 PM Assignment - Jupyter Notebook

Out[297]:

( Time on App Time on Website Length of Membership Yearly Amount Spent \

235 11.08 37.96 4.72 517.17
11 12.6 37.37 3.47 501.93
55 12.36 38.04 3.31 468.91
473 11.47 35.68 1.81 374.27
201 12.52 37.15 2.67 487.38
.. ... ... ... ...
439 13.15 36.62 2.49 470.45
339 11.56 35.98 1.48 282.47
432 12.7 35.36 4.0 553.6
160 11.75 36.94 0.8 298.76
375 12.43 37.63 4.33 532.72

avatar1 avatar2 clc1 clc2 purchase

235 1 0 0 0 1.0
11 0 0 0 0 1.0
55 0 0 1 0 1.0
473 0 0 0 0 0.0
201 0 0 0 0 1.0
.. ... ... ... ... ...
439 0 0 0 0 1.0
339 0 0 0 0 0.0
432 0 0 0 0 1.0
160 0 0 0 0 0.0
375 0 0 0 0 1.0

[400 rows x 9 columns],

Time on App Time on Website Length of Membership Yearly Amount Spent \
97 12.91 36.05 3.49 547.71
496 11.83 36.84 3.61 502.09
57 11.17 35.63 5.46 587.57
334 13.29 38.63 3.87 543.34
95 11.33 35.46 4.54 568.72
.. ... ... ... ...
393 11.54 37.53 2.92 431.62
210 12.05 38.51 2.85 409.09
215 11.67 37.34 4.26 567.48
461 12.5 38.05 4.64 616.85
368 11.41 36.38 4.04 541.05

avatar1 avatar2 clc1 clc2 purchase

97 0 0 0 0 1.0
496 0 0 0 0 1.0
57 0 0 0 0 1.0
334 0 0 0 0 1.0
95 0 0 0 0 1.0
.. ... ... ... ... ...
393 0 0 0 0 0.0
210 0 0 0 0 0.0
215 0 0 0 0 1.0
461 0 0 0 0 1.0
368 0 0 0 0 1.0

[100 rows x 9 columns])

localhost:8888/notebooks/Desktop/Personal/code/Practice_ML/Assignment.ipynb# 8/10
7/28/23, 1:10 PM Assignment - Jupyter Notebook

In [305]:

1 def resample(dataframe, oversample=False):

2 x = dataframe[dataframe.columns[:-1]].values
3 y = dataframe[dataframe.columns[-1]].astype('int').values
4
5 if oversample:
6 ros = RandomOverSampler()
7 x,y = ros.fit_resample(x,y)
8
9 data = np.hstack((x, np.reshape(y,(-1,1))))
10 return data, x, y

In [306]:

1 train, X_train, y_train = resample(train, oversample = True)

2 test, X_test, y_test = resample(test, oversample = False)

In [332]:

1 from sklearn.linear_model import LogisticRegression

2
3 lg_model = LogisticRegression(solver='lbfgs', max_iter=100)
4 lg_model.fit(X_train, y_train)

C:\Users\creat\anaconda3\lib\site-packages\sklearn\linear_model\_logistic.py:458: Co
nvergenceWarning: lbfgs failed to converge (status=1):
STOP: TOTAL NO. of ITERATIONS REACHED LIMIT.

Increase the number of iterations (max_iter) or scale the data as shown in:
https://scikit-learn.org/stable/modules/preprocessing.html (https://scikit-lear
n.org/stable/modules/preprocessing.html)
Please also refer to the documentation for alternative solver options:
https://scikit-learn.org/stable/modules/linear_model.html#logistic-regression (h
ttps://scikit-learn.org/stable/modules/linear_model.html#logistic-regression)
n_iter_i = _check_optimize_result(

Out[332]:

▾ LogisticRegression
LogisticRegression()

In [333]:

1 y_pred = lg_model.predict(X_test)

In [334]:

1 from sklearn.metrics import classification_report

In [335]:

1 print(classification_report(y_test, y_pred))

precision recall f1-score support

0 0.93 1.00 0.96 26

1 1.00 0.97 0.99 74

accuracy 0.98 100

macro avg 0.96 0.99 0.97 100
weighted avg 0.98 0.98 0.98 100

localhost:8888/notebooks/Desktop/Personal/code/Practice_ML/Assignment.ipynb# 9/10
7/28/23, 1:10 PM Assignment - Jupyter Notebook

Summary
1. Created and cleaned address to get relaitable data point for the users to see similarity in behaviour-- built 2 more
colunms -- cleaned_address_loc and cleaned_address_county
2. Load the data and cleaned the columns name
3. Built the purchase colunms based on the contition mentioned
4. Looked on the basic stats before cleaning and resampling the data (describe and info)
5. Vectorize clean_address_county and avatar to use them in the regression
6. Split into train and test as mentioned (80:20)
7. Resample -- oversample to have a decent data to make generic model
8. build logistic regression model and predict using the same on the test data
9. Showcase the stats [ F1 : 98% accuracy ]

Insights
1. Higher the time spend on App and lenght of membership -- higher the probaboility they will make a purchase
2. App is more efective then web for purchase conversion
3. Lenght of membership has highest impact on the purchase

localhost:8888/notebooks/Desktop/Personal/code/Practice_ML/Assignment.ipynb# 10/10