-
In this project, a clean and transformed 'Mall_Customers Dataset' obtained from a shopping mall's database is presented.
-
My aim is to cluster the given dataset into distinct target market groups and analyze the obtained target groups.This is in a bid to gain insight on target markets that may potentially lead to increased product sales at the mall when advertised to.
-
I shall carry out my analysis using the 'Python Programming Language' on the Jupyter Notebook IDE.
-
I first imported libraries that shall be useful in performing my analysis.
-
Pandas for data manipulation & data wrangling,Seaborn for statistical visualization,matplotlib.pyplot for data visualization,sklearn.cluster for clustering and warnings for removing py prompt warnings.
-
Code: import pandas as pd,import seaborn as sns,import matplotlib.pyplot as plt,from sklearn.cluster import KMeans,import warnings warnings.filterwarnings('ignore')
-
-
I then proceeded to import the Mall_Customers dataset(named it 'df') and then confirmed that it loaded correctly by running a data overview query.
- Code 1: df = pd.read_csv() and Code 2: df.head()
-
Checked for null values across all the columns.No column in this dataset contains null values (no null values).
- Code: pd.isnull(df).sum()
-
Obtained information about the dataset.The dataset has a total of 5 columns and 200 records.In addittion, by running the code, I was able to know the column names and the data types of data values in each column.
-
Code: df.info()
-
Output:
-
Column Non-Null Count Dtype
0 CustomerID 200 non-null int64
1 Gender 200 non-null object
2 Age 200 non-null int64
3 Annual Income (k$) 200 non-null int64
4 Spending Score (1-100) 200 non-null int64
dtypes: int64(4), object(1)
-
1). For starters,I obtained the descriptive statistics of the dataset variables.
-
Code: df.describe().transpose()
-
Output: count mean std min 25% 50% 75% max
CustomerID 200.0 100.50 57.879185 1.0 50.75 100.5 150.25 200.0
Age 200.0 38.85 13.969007 18.0 28.75 36.0 49.00 70.0
Annual Income (k$) 200.0 60.56 26.264721 15.0 41.50 61.5 78.00 137.0
Spending Score (1-100) 200.0 50.20 25.823522 1.0 34.75 50.0 73.00 99.0
2). Histogram plots for each of the numerical variables.
-
Code:
columns=['CustomerID', 'Age', 'Annual Income (k$)']
for i in columns:
plt.figure()
sns.distplot(df[i])
-
Insights:
-
The 'CustomerID' Histogram confirms and displays that our dataset contains 200 customers.
-
From the 'Annual Income' Histogram, I observed that the age range of the customers is between 18-69 whereby most customers are in the age range 18-52.
-
From the 'Annual Income' Histogram,I observed that most of the customers lie between the 50k-80k annual income range while the total range lies between 18k to 52k.
-
3). KDE Plots for each of the numerical variables,categorized by gender.
-
Code:
columns = ['Age', 'Annual Income (k$)','Spending Score (1-100)']
for i in columns:
plt.figure()
sns.kdeplot(df[i],shade=True,hue=df['Gender'])
-
Insights:
-
From the Ist visualization, the female customers are more in number within the age range of about 18 to 58.In the 0-18 and 58-80 age ranges,the male gender customers are more in number.
-
In the 2nd visualization, the female customers earn more annual income than the male customers within the income range 0-130k while in the 130-150k range the male customers outshine the female customers.
-
Generally, female customers have a significantly higher spending score as compared to their male counterparts.
-
4). Boxplots for each numerical variable(categorized by gender).
-
Code:
columns = ['Age', 'Annual Income (k$)','Spending Score (1-100)']
for i in columns:
plt.figure()
sns.boxplot(data=df,x='Gender',y=df[i])
-
From the boxplots,we can observe that there are barely any outliers in the dataset variables.
5). Number of male and female customers in % form.
-
Code: df['Gender'].value_counts(normalize=True)
-
It is clear to see that the female customers outnumber the male customers by 12%.
-
Output:
Female 0.56
Male 0.44
Name: Gender, dtype: float64
1). Scatter plot to show the relationship between 'Annual Income and 'Spending score'.
-
From the visualization, I observed that the two variables have a non-linear realtionship.
-
Code: sns.scatterplot(data=df, x='Annual Income (k$)',y='Spending Score (1-100)' )
2) Pairplots for the dataset variables.
-
From the pairplots, I observed the type of relationship each variable has with each other.
-
Code:
df=df.drop('CustomerID',axis=1)
sns.pairplot(df,hue='Gender')
3) Correlation between the variables.
-
From the output, I observed that the 'Age' variable is negatively correlated to the 'Annual income' and 'Spending score' variables.
-
While the 'Annual income' is positively correlated to the 'Spending score'.
-
Code: df.corr()
-
Output:
Age Annual Income (k$) Spending Score (1-100) Age 1.000000 -0.012398 -0.327227 Annual Income (k$) -0.012398 1.000000 0.009903 Spending Score (1-100) -0.327227 0.009903 1.000000
4). Mean values of the numerical data variables,grouped by gender.
-
By observation, although the male customers have an averagely higher income though they spend less in comparision to the female customers.
-
Code: df.groupby(['Gender'])['Age', 'Annual Income (k$)','Spending Score (1-100)'].mean()
-
Output:
Age Annual Income (k$) Spending Score (1-100) Gender Female 38.098214 59.250000 51.526786 Male 39.806818 62.227273 48.511364
5). Heatmap to display the correlation between variables.
-
To read the visualization, -1>x>0 indicates a negative correlatio between the variables while 0>x<1 indicates a positive correlation between the variables.
i) For loop to determine the intertia scores for clusters between the range of 1-11.
-
Code:
intertia_scores=[]
for i in range(1,11):
kmeans=KMeans(n_clusters=i) kmeans.fit(df[['Annual Income (k$)']]) intertia_scores.append(kmeans.inertia_)
ii) Plot the intertia scores to obtain the most suitable cluster number,this is through the elbow method.
-
From the plot, I determined the appropriate lot size to be 3.
-
Code: plt.plot(range(1,11),intertia_scores)
iii) Initiate my algorithm.
- Code: clustering1 = KMeans(n_clusters=3)
iv) Fit my algorithm into the data.
- Code: clustering1.fit(df[['Annual Income (k$)']])
v) Add an 'Income Cluster' column that contains clustered data values to the existing dataset.
-
Code 1: df['Income Cluster'] = clustering1.labels and Code 2: df.head()
-
Output:
Gender Age Annual Income (k$) Spending Score (1-100) Income Cluster0 Male 19 15 39 0
1 Male 21 15 81 0
2 Female 20 16 6 0
3 Female 23 16 77 0
4 Female 31 17 40 0
vi) Obtain the number of data values in each cluster.
-
Code: df['Income Cluster'].value_counts()
-
Output:
- The **'1' cluster** has the largest number of customers i.e **92**. - The **'0' cluster** has the second largest number of customers i.e **72**. - The **'2'** cluster has the least number of customers i.e **36**.
vii) Summary statistics of the clusters obtained.
-
The income cluster 2 has the highest level of annual income in addittion to the highest spending score.
-
Code: df.groupby('Income Cluster')['Age', 'Annual Income (k$)','Spending Score (1-100)'].mean()
-
Output:
Age Annual Income (k$) Spending Score (1-100) Income Cluster 0 38.930556 33.027778 50.166667 1 39.184783 66.717391 50.054348 2 37.833333 99.888889 50.638889
i). Initiate the KMeans clustering algorithm,fit the data values,label the obtained clusters, create new column,obtain dataset overview.
-
Code:
clustering2 = KMeans(n_clusters=5)
clustering2.fit(df[['Annual Income (k$)','Spending Score (1-100)']])
df['Spending and Income Cluster'] =clustering2.labels_
df.head()
-
Output:
Gender Age Annual Income (k$) Spending Score (1-100) Income Cluster Spending and Income Cluster 0 Male 19 15 39 0 4 1 Male 21 15 81 0 3 2 Female 20 16 6 0 4 3 Female 23 16 77 0 3 4 Female 31 17 40 0 4
ii). Obtain the appropriate number of clusters for the dataset using the elbow method.
-
From the plot,I was able to obtain that the suitable number of clusters is 5.
-
Code:
intertia_scores2=[]
for i in range(1,11):
kmeans2=KMeans(n_clusters=i)
kmeans2.fit(df[['Annual Income (k$)','Spending Score (1-100)']])
intertia_scores2.append(kmeans2.inertia_)
plt.plot(range(1,11),intertia_scores2)
iii). Scatter plot to display the relationship between the 'Spending score' and 'Annual Income' in reference to clusters.
-
Code 1:
centers =pd.DataFrame(clustering2.cluster_centers_)
centers.columns = ['x','y']
-
Code 2:
plt.figure(figsize=(10,8))
plt.scatter(x=centers['x'],y=centers['y'],s=100,c='black',marker='*')
sns.scatterplot(data=df, x ='Annual Income (k$)',y='Spending Score (1-100)',hue='Spending and Income Cluster',palette='tab10')
plt.savefig('clustering_bivaraiate.png')
iv) Compare the male and female gender by the 'Spending and Income Cluster'.
-
Code: pd.crosstab(df['Spending and Income Cluster'],df['Gender'],normalize='index')
-
Output:
Gender Female Male
Spending and Income Cluster
0 0.592593 0.407407
1 0.538462 0.461538
2 0.457143 0.542857
3 0.590909 0.409091
4 0.608696 0.391304
v). Group the variables by the 'Spending and Income' Cluster.
-
Code: df.groupby('Spending and Income Cluster')['Age', 'Annual Income (k$)','Spending Score (1-100)'].mean()
-
Output:
Age Annual Income (k$) Spending Score (1-100)Spending and Income Cluster
0 42.716049 55.296296 49.518519 1 32.692308 86.538462 82.128205 2 41.114286 88.200000 17.114286 3 25.272727 25.727273 79.363636 4 45.217391 26.304348 20.913043
-
The aim of the project is to identify the best target customer group(s) in order for the marketing team to plan the best optimal strategy that ought to increase the mall's product sales.
-
Following my analysis above,my reccommendations are as follows:
1). The best target customer group is: Cluster 1.
-
This is because this cluster has both a high spending score and high income thus their able to purchase more products and at an often rate too.
-
54% of cluster 1 customers are women.The marketing team could run a marketing campaign advertising items that particularly resonate with this sub-cluster.
2) Cluster 3 also presents an interesting opportunity to increase product sales.
-
Although this cluster is in the low income range,it is suprisingly also in the high spending score category.
-
The customers could potentially be inclined to purchase more products when popular items are discounted.
-