This project involves analyzing customer data from a New York City clothing store that sells products both online and in person. Customers often receive in-store advice and later make purchases through the mobile app or website. The company aims to determine whether to focus on improving the mobile app experience or the website.

The dataset contains the following columns:

• Avg. Session Length: Average duration of in-store style advice sessions.
• Time on App: Average time spent on the app in minutes.
• Time on Website: Average time spent on the website in minutes.
• Length of Membership: Number of years the customer has been a member.
• Yearly Amount Spent: Total dollars spent by the customer annually.

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline

customers = pd.read_csv("Ecommerce Customers.csv")

customers.head()
customers.info()
customers.describe()

We will focus on the numerical data to understand the relationships between different variables.

sns.set_palette("GnBu_d")
sns.set_style('whitegrid')
sns.jointplot(x='Time on Website', y='Yearly Amount Spent', data=customers)
sns.jointplot(x='Time on App', y='Yearly Amount Spent', data=customers)
sns.jointplot(x='Time on App', y='Length of Membership', kind='hex', data=customers)

sns.pairplot(customers)

sns.lmplot(x='Length of Membership', y='Yearly Amount Spent', data=customers)

We split the data into training and testing sets to evaluate our model.

from sklearn.model_selection import train_test_split

X = customers[['Avg. Session Length', 'Time on App', 'Time on Website', 'Length of Membership']]
y = customers['Yearly Amount Spent']

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=101)

We use a Linear Regression model to fit our training data.

from sklearn.linear_model import LinearRegression

lm = LinearRegression()
lm.fit(X_train, y_train)
print('Coefficients: \n', lm.coef_)

We evaluate the model's performance using the test data.

predictions = lm.predict(X_test)

plt.scatter(y_test, predictions)
plt.xlabel('Y Test')
plt.ylabel('Predicted Y')

We calculate various metrics to evaluate the model's performance.

from sklearn import metrics

print('MAE:', metrics.mean_absolute_error(y_test, predictions))
print('MSE:', metrics.mean_squared_error(y_test, predictions))
print('RMSE:', np.sqrt(metrics.mean_squared_error(y_test, predictions)))

sns.distplot((y_test - predictions), bins=50)

To decide whether to focus on the mobile app or the website, we interpret the coefficients of the model.

coefficients = pd.DataFrame(lm.coef_, X.columns)
coefficients.columns = ['Coefficient']
coefficients

• Avg. Session Length: A 1 unit increase is associated with a $25.98 increase in yearly spending.
• Time on App: A 1 unit increase is associated with a $38.59 increase in yearly spending.
• Time on Website: A 1 unit increase is associated with a $0.19 increase in yearly spending.
• Length of Membership: A 1 unit increase is associated with a $61.27 increase in yearly spending.

While both the app and website contribute to yearly spending, the app shows a stronger correlation. However, further analysis of the relationship between Length of Membership and the app/website could provide additional insights.

For detailed code and further analysis, refer to the project notebook.