EDA Workflow — SegmentIQ

1

Dataset Shape

The first thing we check is the number of rows (customers) and columns (features). This tells us the scale of the dataset and helps decide if we need sampling.

              Python
               import pandas as pd

               df = pd.read_csv('Mall_Customers.csv')

               print(df.shape)  # Output: (rows, columns)

Result: Shape = 200 rows × 5 columns — 200 customers, 5 features.

2

Column Names

List all column names to understand what data is available before analysis.

              Python
               print(df.columns.tolist())
            

{'Feature': 'Age', 'count': 200.0, 'mean': 38.85, 'std': 13.97, 'min': 18.0, '25%': 28.75, '50%': 36.0, '75%': 49.0, 'max': 70.0} {'Feature': 'Annual_Income', 'count': 200.0, 'mean': 60.56, 'std': 26.26, 'min': 15.0, '25%': 41.5, '50%': 61.5, '75%': 78.0, 'max': 137.0} {'Feature': 'Spending_Score', 'count': 200.0, 'mean': 50.2, 'std': 25.82, 'min': 1.0, '25%': 34.75, '50%': 50.0, '75%': 73.0, 'max': 99.0} {'Feature': 'Gender_Encoded', 'count': 200.0, 'mean': 0.44, 'std': 0.5, 'min': 0.0, '25%': 0.0, '50%': 0.0, '75%': 1.0, 'max': 1.0}

3

Data Types

Verifying data types ensures that numeric columns aren't stored as strings and that categorical columns are correctly typed.

              Python
               print(df.dtypes)

               print(df.info())
            

Column	Type	Sample Value
Gender	str	Male
Age	int64	19
Annual_Income	int64	15
Spending_Score	int64	39
Gender_Encoded	int64	1

4

Missing Values Check

Missing values can cause K-Means to fail or produce incorrect clusters. We check each column for nulls before proceeding.

              Python
               print(df.isnull().sum())  # Count nulls per column

               print(df.isnull().sum().sum())  # Total nulls
            

Column	%	Status
Gender	0.0%	✅ Clean
Age	0.0%	✅ Clean
Annual_Income	0.0%	✅ Clean
Spending_Score	0.0%	✅ Clean
Gender_Encoded	0.0%	✅ Clean

✅ No missing values detected. All 200 rows are complete.

5

Duplicate Rows

Duplicate rows skew clustering results by creating artificial density in certain regions of the feature space.

              Python
               print(df.duplicated().sum())

               df = df.drop_duplicates()
            

Duplicates found: 0 — No duplicates, dataset is clean.

6

Statistical Summary

df.describe() gives count, mean, std, min/max, and quartiles for each numeric column.

              Python
               df.describe().round(2)
            

Feature	Count	Mean	Std	Min	25%	50%	75%	Max
Age	200	38.85	13.97	18.0	28.75	36.0	49.0	70.0
Annual_Income	200	60.56	26.26	15.0	41.5	61.5	78.0	137.0
Spending_Score	200	50.2	25.82	1.0	34.75	50.0	73.0	99.0
Gender_Encoded	200	0.44	0.5	0.0	0.0	0.0	1.0	1.0

7

Outlier Detection (IQR Method)

The Inter-Quartile Range (IQR) method detects outliers: values below Q1−1.5×IQR or above Q3+1.5×IQR.

              Python
               Q1, Q3 = df[col].quantile(0.25), df[col].quantile(0.75)

               IQR = Q3 - Q1  # Inter-Quartile Range

               outliers = df[(df[col] < Q1-1.5*IQR) | (df[col] > Q3+1.5*IQR)]

Feature	Q1	Q3	IQR	Lower Fence	Upper Fence	Outliers
Age	28.75	49.0	20.25	-1.62	79.38	0
Annual_Income	41.5	78.0	36.5	-13.25	132.75	2
Spending_Score	34.75	73.0	38.25	-22.62	130.38	0

8

Gender Distribution

👩

112

Female

👨

88

Male

9

Feature Value Ranges

Feature	Min	Max	Mean	Median	Std
Age	18.0	70.0	38.85	36.0	13.97
Annual_Income	15.0	137.0	60.56	61.5	26.26
Spending_Score	1.0	99.0	50.2	50.0	25.82
Gender_Encoded	0.0	1.0	0.44	0.0	0.5

10

Correlation Analysis

Pearson correlation measures linear relationships (−1 to +1). Values near 0 indicate independent features — good for K-Means as we want feature diversity.

Feature	Age	Annual_Income	Spending_Score	Gender_Encoded
Age	1.0	-0.012	-0.327	0.061
Annual_Income	-0.012	1.0	0.01	0.056
Spending_Score	-0.327	0.01	1.0	-0.058
Gender_Encoded	0.061	0.056	-0.058	1.0

Low correlation between Income and Spending Score (key features) confirms they capture independent dimensions of customer behaviour — ideal for clustering.

🔍 Step-by-Step EDA Workflow