Puma US Sales Analysis¶
Data Overview¶
In [2]:
import pandas as pd
df = pd.read_csv(r"C:\Users\lakshita rawat\Downloads\Project_2\Dataset_1.csv")
df.head()
Out[2]:
| Retailer | Months | Year | Invoice Date | Region | State | City | Product | Price per Unit($) | Units Sold | Total Sales($) | Operating Profit($) | Operating Margin | Sales Method | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | Foot Locker | January | 2020 | 01-01-2020 | Northeast | New York | New York | Men's Street Footwear | 50 | 1200 | 600000 | 300000 | 0.50 | In-store |
| 1 | Foot Locker | January | 2020 | 02-01-2020 | Northeast | New York | New York | Men's Athletic Footwear | 50 | 1000 | 500000 | 150000 | 0.30 | In-store |
| 2 | Foot Locker | January | 2020 | 03-01-2020 | Northeast | New York | New York | Women's Street Footwear | 40 | 1000 | 400000 | 140000 | 0.35 | In-store |
| 3 | Foot Locker | January | 2020 | 04-01-2020 | Northeast | New York | New York | Women's Athletic Footwear | 45 | 850 | 382500 | 133875 | 0.35 | In-store |
| 4 | Foot Locker | January | 2020 | 05-01-2020 | Northeast | New York | New York | Men's Apparel | 60 | 900 | 540000 | 162000 | 0.30 | In-store |
A quick overview of the dataset helps us understand its structure, key variables, and basic statistics, ensuring the data is ready for deeper analysis¶
In [ ]:
print(df.shape) #total rows and columns
print()
print(df.columns)
print()
print(df.isnull().sum())
print()
print(df.duplicated().sum())
print()
print(df.describe())
print()
print(df.info())
(9648, 14)
Index(['Retailer', 'Months', 'Year', 'Invoice Date', 'Region', 'State', 'City',
'Product', 'Price per Unit($)', 'Units Sold', 'Total Sales($)',
'Operating Profit($)', 'Operating Margin', 'Sales Method'],
dtype='object')
Retailer 0
Months 0
Year 0
Invoice Date 0
Region 0
State 0
City 0
Product 0
Price per Unit($) 0
Units Sold 0
Total Sales($) 0
Operating Profit($) 0
Operating Margin 0
Sales Method 0
dtype: int64
0
Year Price per Unit($) Units Sold Total Sales($) \
count 9648.000000 9648.000000 9648.000000 9648.000000
mean 2020.865050 45.216625 256.930037 93273.437500
std 0.341688 14.705397 214.252030 141916.016727
min 2020.000000 7.000000 0.000000 0.000000
25% 2021.000000 35.000000 106.000000 4254.500000
50% 2021.000000 45.000000 176.000000 9576.000000
75% 2021.000000 55.000000 350.000000 150000.000000
max 2021.000000 110.000000 1275.000000 825000.000000
Operating Profit($) Operating Margin
count 9648.000000 9648.000000
mean 34425.282131 0.422991
std 54193.124141 0.097197
min 0.000000 0.100000
25% 1922.000000 0.350000
50% 4371.500000 0.410000
75% 52063.000000 0.490000
max 390000.000000 0.800000
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 9648 entries, 0 to 9647
Data columns (total 14 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 Retailer 9648 non-null object
1 Months 9648 non-null object
2 Year 9648 non-null int64
3 Invoice Date 9648 non-null object
4 Region 9648 non-null object
5 State 9648 non-null object
6 City 9648 non-null object
7 Product 9648 non-null object
8 Price per Unit($) 9648 non-null int64
9 Units Sold 9648 non-null int64
10 Total Sales($) 9648 non-null int64
11 Operating Profit($) 9648 non-null int64
12 Operating Margin 9648 non-null float64
13 Sales Method 9648 non-null object
dtypes: float64(1), int64(5), object(8)
memory usage: 1.0+ MB
None
In [16]:
# All unique states
unique_states = df['State'].unique()
print(unique_states)
print()
# Total Count
num_states = df['State'].nunique()
print("Number of unique states:", num_states)
print()
# All unique cities
unique_cities = df['City'].unique()
print(unique_cities)
print()
# Total Count
num_states = df['City'].nunique()
print("Number of unique cities:", num_states)
['New York' 'Texas' 'California' 'Illinois' 'Pennsylvania' 'Nevada' 'Colorado' 'Washington' 'Florida' 'Minnesota' 'Montana' 'Tennessee' 'Nebraska' 'Alabama' 'Maine' 'Alaska' 'Hawaii' 'Wyoming' 'Virginia' 'Michigan' 'Missouri' 'Utah' 'Oregon' 'Louisiana' 'Idaho' 'Arizona' 'New Mexico' 'Georgia' 'South Carolina' 'North Carolina' 'Ohio' 'Kentucky' 'Mississippi' 'Arkansas' 'Oklahoma' 'Kansas' 'South Dakota' 'North Dakota' 'Iowa' 'Wisconsin' 'Indiana' 'West Virginia' 'Maryland' 'Delaware' 'New Jersey' 'Connecticut' 'Rhode Island' 'Massachusetts' 'Vermont' 'New Hampshire'] Number of unique states: 50 ['New York' 'Houston' 'San Francisco' 'Los Angeles' 'Chicago' 'Dallas' 'Philadelphia' 'Las Vegas' 'Denver' 'Seattle' 'Miami' 'Minneapolis' 'Billings' 'Knoxville' 'Omaha' 'Birmingham' 'Portland' 'Anchorage' 'Honolulu' 'Orlando' 'Albany' 'Cheyenne' 'Richmond' 'Detroit' 'St. Louis' 'Salt Lake City' 'New Orleans' 'Boise' 'Phoenix' 'Albuquerque' 'Atlanta' 'Charleston' 'Charlotte' 'Columbus' 'Louisville' 'Jackson' 'Little Rock' 'Oklahoma City' 'Wichita' 'Sioux Falls' 'Fargo' 'Des Moines' 'Milwaukee' 'Indianapolis' 'Baltimore' 'Wilmington' 'Newark' 'Hartford' 'Providence' 'Boston' 'Burlington' 'Manchester'] Number of unique cities: 52
In [39]:
df.columns
Out[39]:
Index(['Retailer', 'Months', 'Year', 'Invoice Date', 'Region', 'State', 'City',
'Product', 'Price per Unit($)', 'Units Sold', 'Total Sales($)',
'Operating Profit($)', 'Operating Margin', 'Sales Method'],
dtype='object')
- The overall distribution of total sales is analyzed to understand how sales are spread across the dataset.
- The analysis identifies which products contribute the most to overall sales.
- Sales are studied across different regions, states, and cities to capture the geographical distribution.
- The distribution of units sold is examined to highlight sales volume patterns across products.
- The sales methods (online, offline, wholesale, etc.) are explored to determine the most frequently used channel.
In [ ]:
import matplotlib.pyplot as plt
# 1. Total Sales by Product
sales_by_product = df.groupby('Product')['Total Sales($)'].sum().sort_values(ascending=False)
ax = sales_by_product.plot(kind='bar', figsize=(10,6), color="skyblue")
for i, v in enumerate(sales_by_product):
ax.text(i, v, f'{int(v)}', ha='center', va='bottom')
plt.title("Total Sales by Product")
plt.ylabel("Total Sales ($)")
plt.show()
# 2. Units Sold by Product
units_by_product = df.groupby('Product')['Units Sold'].sum().sort_values(ascending=False)
ax = units_by_product.plot(kind='bar', figsize=(10,6), color="lightgreen")
for i, v in enumerate(units_by_product):
ax.text(i, v, f'{int(v)}', ha='center', va='bottom')
plt.title("Total Units Sold by Product")
plt.ylabel("Units Sold")
plt.show()
# 3. Total Sales by Region
sales_by_region = df.groupby('Region')['Total Sales($)'].sum().sort_values(ascending=False)
ax = sales_by_region.plot(kind='bar', figsize=(8,6), color="salmon")
for i, v in enumerate(sales_by_region):
ax.text(i, v, f'{int(v)}', ha='center', va='bottom')
plt.title("Total Sales by Region")
plt.ylabel("Total Sales ($)")
plt.show()
# 4. Sales Count by Method
method_counts = df['Sales Method'].value_counts()
ax = method_counts.plot(kind='bar', figsize=(6,5), color="orange")
for i, v in enumerate(method_counts):
ax.text(i, v, str(v), ha='center', va='bottom')
plt.title("Number of Transactions by Sales Method")
plt.ylabel("Count")
plt.show()
# 5. Total Sales by Retailer
sales_by_retailer = df.groupby('Retailer')['Total Sales($)'].sum().sort_values(ascending=False)
ax = sales_by_retailer.plot(kind='bar', figsize=(12,6), color="purple")
for i, v in enumerate(sales_by_retailer):
ax.text(i, v, f'{int(v)}', ha='center', va='bottom', fontsize=9)
plt.title("Total Sales by Retailer")
plt.ylabel("Total Sales ($)")
plt.show()
Time Analysis (Year, Month, Invoice Date)¶
Here through the analysis we'll get to know:
- Is Puma’s sales revenue growing, stable, or declining year-over-year?
- Which months consistently show peak sales, and which months lag behind?
- How do monthly sales trends compare across different years?
- Is the AOV growing, suggesting stronger brand positioning, or shrinking, indicating price sensitivity?
- How are online, offline, and in-store sales channels performing over time?
In [41]:
df.columns
Out[41]:
Index(['Retailer', 'Months', 'Year', 'Invoice Date', 'Region', 'State', 'City',
'Product', 'Price per Unit($)', 'Units Sold', 'Total Sales($)',
'Operating Profit($)', 'Operating Margin', 'Sales Method'],
dtype='object')
In [ ]:
import pandas as pd
import matplotlib.pyplot as plt
# 1. Year-over-Year Sales Trend
yearly_sales = df.groupby('Year')['Total Sales($)'].sum()
ax = yearly_sales.plot(kind='bar', color="skyblue", figsize=(8,5))
for i, v in enumerate(yearly_sales):
ax.text(i, v, f'{int(v)}', ha='center', va='bottom')
plt.title("Year-over-Year Sales Revenue")
plt.ylabel("Total Sales ($)")
plt.show()
# 2. Monthly Sales (Across All Years) – Seasonality
monthly_sales = df.groupby('Months')['Total Sales($)'].sum()
ax = monthly_sales.plot(kind='bar', color="lightgreen", figsize=(10,5))
for i, v in enumerate(monthly_sales):
ax.text(i-0.1, v, f'{int(v)}', ha='center', va='bottom')
plt.title("Total Sales by Month (Seasonality)")
plt.xlabel("Months")
plt.ylabel("Total Sales ($)")
plt.show()
# 3. Monthly Sales Trend across Years with Annotations
monthly_trend = df.groupby(['Year','Months'])['Total Sales($)'].sum().unstack(0)
ax = monthly_trend.plot(marker='o', figsize=(12,6))
# Add annotations
for year in monthly_trend.columns:
for i, v in enumerate(monthly_trend[year]):
if pd.notna(v):
ax.text(i+1, v, f'{int(v):,}', ha='center', va='bottom', fontsize=8, rotation=0)
plt.title("Monthly Sales Trend Across Years")
plt.xlabel("Months")
plt.ylabel("Total Sales ($)")
plt.xticks(range(1,13), ['Jan','Feb','Mar','Apr','May','Jun','Jul','Aug','Sep','Oct','Nov','Dec'])
plt.legend(title="Year")
plt.grid(True, linestyle="--", alpha=0.6)
plt.show()
# 4. Average Order Value (AOV) per Month
monthly_orders = df.groupby(['Year','Months']).agg(
total_sales=('Total Sales($)','sum'),
total_orders=('Invoice Date','count')
)
monthly_orders['AOV'] = monthly_orders['total_sales'] / monthly_orders['total_orders']
aov = monthly_orders['AOV'].unstack(0)
ax = aov.plot(kind='line', marker='o', figsize=(12,6))
# Add values above points
for year in aov.columns:
for i, v in enumerate(aov[year]):
ax.text(i, v, f'{v:.2f}', ha='center', va='bottom')
plt.title("Average Order Value (AOV) per Month")
plt.ylabel("AOV ($)")
plt.xlabel("Month")
plt.xticks(range(1,13), ['Jan','Feb','Mar','Apr','May','Jun','Jul','Aug','Sep','Oct','Nov','Dec'])
plt.grid(True, linestyle="--", alpha=0.6)
plt.show()
# 5. Yearly Sales Method Growth
# Ensure Year column is integer (sometimes it's string/object)
df['Year'] = df['Year'].astype(int)
# Group by Year and Sales Method
method_trend = df.groupby(['Year', 'Sales Method'])['Total Sales($)'].sum().reset_index()
# Pivot for plotting (Years as x-axis, Methods as columns)
pivot_trend = method_trend.pivot(index='Year', columns='Sales Method', values='Total Sales($)')
# Line chart
ax = pivot_trend.plot(kind='line', marker='o', figsize=(10,6))
# Add values above each point
for method in pivot_trend.columns:
for i, v in enumerate(pivot_trend[method]):
if pd.notna(v):
ax.text(pivot_trend.index[i], v, f'{int(v):,}', ha='center', va='bottom', fontsize=9)
plt.title("Sales by Method Over Time")
plt.ylabel("Total Sales ($)")
plt.xlabel("Year")
plt.xticks(pivot_trend.index) # ensure all years are shown
plt.legend(title="Sales Method")
plt.grid(True, linestyle="--", alpha=0.6)
plt.show()
Insights from the graph¶
Year-over-Year Sales Revenue- Sales revenue showed explosive growth from 2020 to 2021, jumping from $182M to $717M, highlighting a strong year-over-year recovery and expansion.
Monthly Sales Seasonality-
- Sales displayed clear seasonal fluctuations.
- July (≈$95M) and August (≈$92M) were peak months, likely driven by summer demand or mid-year promotions,February (≈$61M) and March (≈$57M) marked the lowest sales, suggesting weaker demand in early spring.
- Sales show a clear seasonal pattern, with sharp spikes in July and August, likely boosted by back-to-school demand and heightened sports enthusiasm during the Olympic period in both 2020 and 2021. In contrast, February and March record noticeable dips, reflecting the post-holiday lull, fewer retail events, and transitional slowdown in consumer spending
- Average Order Value (AOV) per Month
- In 2020, AOV was much higher, peaking at $190K+ in September and December, though with high fluctuations month-to-month.
- In 2021, AOV stabilized at lower levels (≈$70K–$80K), indicating a strategy shift toward higher sales volume with smaller basket sizes, possibly due to discounts or entry-level product pushes.
- Sales by Method Over Time
- In-store sales remained the leader, growing steadily from $93M (2020) to $262M (2021).
- Online sales skyrocketed from just $4.5M to $243M, showing massive adoption of digital channels.
- Outlet sales also rose strongly from $83M to $211M, indicating expansion through physical retail presence.
Regional Analysis¶
Here through the analysis we'll get to know:
- Analyzing the State Level Sales data
- Which region has the max sales?
- Which region has the max Operating Profit?
- Which region has the max/min operating margin and finding it's efficiecy?
- Which Sales Method is prefered in which region?
- Finding out the top 10 states and cities by Sales and Operating Profit
In [6]:
df.columns
Out[6]:
Index(['Retailer', 'Months', 'Year', 'Invoice Date', 'Region', 'State', 'City',
'Product', 'Price per Unit($)', 'Units Sold', 'Total Sales($)',
'Operating Profit($)', 'Operating Margin', 'Sales Method'],
dtype='object')
In [20]:
import pandas as pd
import plotly.express as px
import us #library is installed
# Aggregate total sales by state from your main dataset df
state_sales_map = df.groupby('State', as_index=False)['Total Sales($)'].sum()
# Clean state names
state_sales_map['State'] = state_sales_map['State'].str.strip()
# Convert state names to abbreviations (CA, TX, NY, etc.)
state_sales_map['State_Abbr'] = state_sales_map['State'].apply(
lambda x: us.states.lookup(x).abbr if us.states.lookup(x) else None
)
# Drop rows where abbreviation could not be found
state_sales_map = state_sales_map.dropna(subset=['State_Abbr'])
# Choropleth Map
fig = px.choropleth(
state_sales_map,
locations='State_Abbr', # use abbreviations
locationmode="USA-states",
color='Total Sales($)',
scope="usa",
color_continuous_scale="Blues",
title="State-Level Sales Map",
hover_name="State",
hover_data={'Total Sales($)':':,.0f'}
)
fig.show()
# 2. Total Sales by Region
region_perf = df.groupby('Region')[['Total Sales($)', 'Operating Profit($)']].sum().sort_values('Total Sales($)', ascending=False)
ax = region_perf['Total Sales($)'].plot(kind='bar', figsize=(8,5), color="skyblue")
for i, v in enumerate(region_perf['Total Sales($)']):
ax.text(i, v, f'{int(v):,}', ha='center', va='bottom')
plt.title("Total Sales by Region")
plt.ylabel("Sales ($)")
plt.show()
# 3. Operating Profit by Region
ax = region_perf['Operating Profit($)'].plot(kind='bar', figsize=(8,5), color="lightgreen")
for i, v in enumerate(region_perf['Operating Profit($)']):
ax.text(i, v, f'{int(v):,}', ha='center', va='bottom')
plt.title("Operating Profit by Region")
plt.ylabel("Profit ($)")
plt.show()
# 4. Average Operating Margin by Region
region_margin = df.groupby('Region')['Operating Margin'].mean().sort_values(ascending=False)
ax = region_margin.plot(kind='bar', figsize=(8,5), color="orange")
for i, v in enumerate(region_margin):
ax.text(i, v, f'{v:.2f}%', ha='center', va='bottom')
plt.title("Average Operating Margin by Region")
plt.ylabel("Operating Margin (%)")
plt.show()
# 5.Sales Method Distribution by Region
region_method = df.groupby(['Region','Sales Method'])['Total Sales($)'].sum().unstack()
ax = region_method.plot(kind='bar', stacked=True, figsize=(10,6))
plt.title("Sales Method Distribution by Region")
plt.ylabel("Total Sales ($)")
plt.legend(title="Sales Method")
plt.show()
# 6. Top 10 States by Sales (Pie/Donut)
state_sales = df.groupby('State')['Total Sales($)'].sum().sort_values(ascending=False).head(10)
plt.figure(figsize=(8,8))
wedges, texts, autotexts = plt.pie(
state_sales,
labels=state_sales.index,
autopct='%1.1f%%',
startangle=140,
wedgeprops={'linewidth':1, 'edgecolor':'white'}
)
# Donut effect
centre_circle = plt.Circle((0,0),0.70,fc='white')
fig = plt.gcf()
fig.gca().add_artist(centre_circle)
plt.title("Top 10 States by Sales (Share %)", fontsize=14)
plt.show()
# 7. Operating Profit by Top 10 States
top_states_profit = df.groupby('State')['Operating Profit($)'].sum().sort_values(ascending=False).head(10)
ax = top_states_profit.plot(kind='bar', figsize=(10,6), color="mediumseagreen")
for i, v in enumerate(top_states_profit):
ax.text(i, v, f'{int(v)}', ha='center', va='bottom', fontsize=9)
plt.title("Top 10 States by Operating Profit")
plt.ylabel("Operating Profit ($)")
plt.xlabel("State")
plt.xticks(rotation=45, ha="right")
plt.show()
# 8. Top 10 Cities by Sales (Pie/Donut)
city_sales = df.groupby('City')['Total Sales($)'].sum().sort_values(ascending=False).head(10)
plt.figure(figsize=(8,8))
wedges, texts, autotexts = plt.pie(
city_sales,
labels=city_sales.index,
autopct='%1.1f%%',
startangle=140,
wedgeprops={'linewidth':1, 'edgecolor':'white'}
)
centre_circle = plt.Circle((0,0),0.70,fc='white')
fig = plt.gcf()
fig.gca().add_artist(centre_circle)
plt.title("Top 10 Cities by Sales (Share %)", fontsize=14)
plt.show()
# 9. Operating Profit by Top 10 Cities
top_cities_profit = df.groupby('City')['Operating Profit($)'].sum().sort_values(ascending=False).head(10)
ax = top_cities_profit.plot(kind='bar', figsize=(10,6), color="cornflowerblue")
for i, v in enumerate(top_cities_profit):
ax.text(i, v, f'{int(v)}', ha='center', va='bottom', fontsize=9)
plt.title("Top 10 Cities by Operating Profit")
plt.ylabel("Operating Profit ($)")
plt.xlabel("City")
plt.xticks(rotation=45, ha="right")
plt.show()
In [ ]:
import matplotlib.pyplot as plt
# Group by Region
region_summary = df.groupby('Region').agg({
'Total Sales($)': 'sum',
'Operating Profit($)': 'sum'
})
region_summary['Operating Margin (%)'] = (region_summary['Operating Profit($)'] / region_summary['Total Sales($)']) * 100
# Plot
fig, ax1 = plt.subplots(figsize=(10,6))
# Bar chart for Sales & Profit
region_summary[['Total Sales($)', 'Operating Profit($)']].plot(
kind='bar', ax=ax1, width=0.7, color=['skyblue','lightgreen']
)
ax1.set_ylabel("Sales / Profit ($)")
ax1.set_title("Regional Comparison: Sales, Profit & Operating Margin")
# Add value labels for bars
for container in ax1.containers:
ax1.bar_label(container, fmt='%d', fontsize=8)
# Second y-axis for Margin
ax2 = ax1.twinx()
ax2.plot(region_summary.index, region_summary['Operating Margin (%)'],
color='red', marker='o', linewidth=2, label="Operating Margin (%)")
ax2.set_ylabel("Operating Margin (%)")
# Add value labels for margin points
for i, v in enumerate(region_summary['Operating Margin (%)']):
ax2.text(i, v, f"{v:.1f}%", ha='center', va='bottom', color='red', fontsize=9)
# Legend
ax1.legend(loc='upper left')
ax2.legend(loc='upper right')
plt.xticks(rotation=45, ha='right')
plt.grid(axis='y', linestyle='--', alpha=0.6)
plt.show()
Insights from the graph¶
The West leads in sales and profit but struggles with the lowest margin, mainly due to higher costs and discount-heavy strategies, while the South, despite lower sales, achieves the highest margin by maintaining tighter cost control and focusing on more profitable sales, it has the healthiest balance between profit and margin. Even though sales are lower than the West, the South earns more per sale, which means sustainable profitability.
The Midwest and Northeast show moderate sales and profits, with margins balancing between West’s low efficiency and South’s high efficiency, indicating steady but less extreme performance.