Customers-Segmentation-Unsupervised-ML-project-in-Python

Problem Statement:

• understand the target customers to plan a marketing strategy.
• identify the most important shopping groups based on annual income, age, gender, and their shopping score.
• identify the ideal number of groups and labeling them using the Elbow's method.

Project approach:

• EDA (Exploratory Data Analysis)
• Utitlize KMeans clustering algorithm
• from the created clusters use Summary Statistics

EDA (Exploratory Data Analysis)

1. Univariate Analysis: to describe and summarize the distribution of a single variable

I) looking at distributions of each column

1. Distribution of Age:

• it's roughly distributed with a peak for people aged around 30-50 years.
• There're also some slightly peaks around 20 and 40 years, indicating other age clusters

2. Distribution of Annual Income (in $k):

• The income is right skewed where few people has higher annual income.

3. Distribution of spending score:

• The distribution is relatively uniform, where most customers have a moderate spending habits.

I recommend: Marketing strategies could target the 30-35 age group, as they are the most common customers. Also, High-income individuals (80k+) might be valuable for premium services.

II) looking at distribution of each column Vs. Gender

• we can see clearly that 'Females' portion is more frequent than that of 'Males'.

• Also there is an outlier in the 'Male' as it has a right thick tail with the distribution skewed to the right.

  • So to investigate this outlier:

  

  Interpretation:

  • The median income for females appears slightly lower than that for males.
  • There’s one outlier among males, indicating a significantly higher income.

2. Bivariate Analysis:

I) understanding customers behavior based on demographic factors like age and gender in relation to income and spending habits.

• In 'Annual Income vs. Spending Score':
  • There may be a pattern found related to annual income and spending score
• In 'Age vs. Spending Score':
  • No strong correlation wes found.
  • Younger individuals may have slightly higher spending scores.
• In 'Age vs. Annual Income' :
  • No correlation was found
  • Income is different across all age groups

II) identify relationships between variables:

• In 'Age vs. Spending Score' :
  • There is a strong negative correlation (-0.33) between age and spending score.
  • Younger individuals tend to have higher spending scores.
• There's no clear correlation between 'Annual Income' and 'Spending Score'
• There's almost no correlation between 'Age' and 'Annual Income'

---

Using K-means clustering ML algorithm, I found:

Trends & Patterns found:

1. target cluster would be cluster 1 which has a high spending score and high annual income.
2. approxiamtley 54% of females in cluster 1 shoppers are females. we should look for ways to attract more of them using a marketing campaign targeting popular items in this cluster.
3. cluster 4 has a potential in sales. so we should make an event on popular items purchased by them.