📦 Amazon Delivery Time Analysis — Why Are Packages Getting Late?

Amazon customers are complaining about late deliveries. As a Data Analyst, you are given delivery data to investigate — Is delivery time really increasing? Which regions are worst? What factors cause delays? 5 real analytical questions solved with Statistics, Python code, and business recommendations. Data Insights par.

🏢 Business Scenario

Problem: Amazon's customer satisfaction scores have dropped 12% in the last quarter. The operations team suspects delivery times have increased beyond the promised 2-day Prime delivery window. Social media complaints about late packages have spiked 40%. The VP of Operations wants a data-driven investigation.

🎯 Hinglish Mein: Amazon ke customers complain kar rahe hain ki packages late aa rahe hain. Prime delivery 2 din mein promise hai but actual 3-4 din lag rahe hain. Operations team ko pata karna hai — kya delivery time sach mein badha hai? Kaunse regions mein sabse zyada delay hai? Kis wajah se late ho raha hai — distance, warehouse load, ya weather? Data Analyst ke roop mein tumhe 5 key questions answer karne hain with statistical proof.

Business Question: Amazon Prime promises 2-day delivery. Has the actual average delivery time increased beyond this SLA (Service Level Agreement)?

🎯 Hinglish: Amazon ka promise hai 2 din mein delivery. But customers bol rahe hain 3-4 din lag rahe hain. Kya yeh complaint sach hai ya sirf perception? Statistical test se prove karo — sample data ka average 2 din se significantly zyada hai ya nahi?

📋 Analysis Approach:

💻 Python Code:

import numpy as np
from scipy import stats

# Delivery data (days)
delivery_days = [4, 1, 3, 5, 2, 2, 4, 2, 5, 1]
promised = 2  # Prime SLA

# Descriptive Stats
print(f"Sample Mean: {np.mean(delivery_days):.1f} days")
print(f"Sample SD:   {np.std(delivery_days, ddof=1):.2f} days")
print(f"Promised:    {promised} days")

# One-Sample T-Test (right-tailed)
t_stat, p_two_tail = stats.ttest_1samp(delivery_days, promised)
p_one_tail = p_two_tail / 2  # Convert to one-tailed

print(f"\nT-statistic: {t_stat:.3f}")
print(f"P-value (one-tailed): {p_one_tail:.4f}")

if p_one_tail < 0.05:
    print("✅ REJECT H₀: Delivery time is significantly > 2 days!")
else:
    print("❌ FAIL TO REJECT: No significant increase.")

# 95% Confidence Interval
ci = stats.t.interval(0.95, df=len(delivery_days)-1,
                     loc=np.mean(delivery_days),
                     scale=stats.sem(delivery_days))
print(f"\n95% CI: {ci[0]:.2f} to {ci[1]:.2f} days")

📊 Results & Interpretation:

Business Question: Are certain regions experiencing worse delivery times than others? Should Amazon allocate more resources to specific regions?

🎯 Hinglish: Kya North region mein delivery late hai but South mein on-time? Ya sab jagah same problem hai? Agar ek specific region mein zyada delay hai toh wahan warehouses add karne chahiye, delivery partners badhane chahiye. ANOVA test se check karo — 4 regions ke means significantly different hain ya nahi.

📋 Analysis Approach:

💻 Python Code:

import pandas as pd
from scipy import stats

# Regional delivery data
north = [4, 2, 5]   # Mean: 3.67
south = [1, 2, 1]   # Mean: 1.33
east  = [3, 4]       # Mean: 3.50
west  = [5, 2]       # Mean: 3.50

# One-Way ANOVA
f_stat, p_value = stats.f_oneway(north, south, east, west)

print(f"F-statistic: {f_stat:.3f}")
print(f"P-value: {p_value:.4f}")

if p_value < 0.05:
    print("✅ REJECT H₀: Regions have significantly different delivery times!")
else:
    print("❌ FAIL TO REJECT: No significant regional difference.")

# Regional summary
regions = {'North': north, 'South': south, 
           'East': east, 'West': west}
for name, data in regions.items():
    print(f"{name}: Mean = {np.mean(data):.2f} days")

📊 Results & Interpretation:

Business Question: Does greater warehouse-to-customer distance directly cause longer delivery times? How strong is this relationship?

🎯 Hinglish: Kya door ke customers ko zyada late delivery hoti hai? Agar distance aur delivery time strongly correlated hain toh Amazon ko regional warehouses build karne chahiye — taaki distance kam ho aur delivery fast ho. Pearson correlation check karo aur regression line banao prediction ke liye.

📋 Analysis Approach:

💻 Python Code:

from scipy import stats

distance = [120, 45, 200, 350, 80, 60, 150, 90, 300, 30]
delivery = [4, 1, 3, 5, 2, 2, 4, 2, 5, 1]

# Pearson Correlation
r, p_val = stats.pearsonr(distance, delivery)
print(f"Pearson r: {r:.3f}")
print(f"P-value: {p_val:.4f}")
print(f"R-squared: {r**2:.3f} ({r**2*100:.1f}% variance explained)")

# Linear Regression
slope, intercept, r_val, p, se = stats.linregress(distance, delivery)
print(f"\nRegression: Delivery = {intercept:.2f} + {slope:.4f} × Distance")

# Prediction: 250 km distance
pred = intercept + slope * 250
print(f"Predicted delivery for 250km: {pred:.1f} days")

📊 Results & Interpretation:

Business Question: Is the current quarter's delivery performance significantly worse than the previous quarter? When did the deterioration begin?

🎯 Hinglish: Pichle quarter (Q2) mein delivery theek thi — average 2.1 din. Is quarter (Q3) mein complaints badh gayi hain. Kya Q3 sach mein Q2 se significantly worse hai ya yeh normal fluctuation hai? Two-sample t-test se compare karo — dono quarters ke means significantly different hain ya nahi.

📋 Analysis Approach:

💻 Python Code:

# Q2 (previous) vs Q3 (current) delivery data
q2_delivery = [2, 2, 1, 3, 2, 2, 1, 2, 3, 2]  # Mean: 2.0
q3_delivery = [4, 1, 3, 5, 2, 2, 4, 2, 5, 1]  # Mean: 2.9

print(f"Q2 Average: {np.mean(q2_delivery):.1f} days")
print(f"Q3 Average: {np.mean(q3_delivery):.1f} days")
print(f"Increase: +{np.mean(q3_delivery)-np.mean(q2_delivery):.1f} days")

# Welch's T-test (independent, unequal variance)
t_stat, p_two = stats.ttest_ind(q3_delivery, q2_delivery, equal_var=False)
p_one = p_two / 2

print(f"\nT-statistic: {t_stat:.3f}")
print(f"P-value (one-tailed): {p_one:.4f}")

if p_one
👤
Jatin Kumar
Data Analyst & Educator
Python, SQL, Power BI aur Excel mein practical tutorials likhta hoon — taaki data analytics seekhna aasan ho. Portfolio: jatinanalytics.co.in
 Portfolio LinkedIn GitHub Kaggle All Articles
 Share:
💬 Comments (0)
💬 Post Comment
Spam/links allowed nahi hain — respectful comments welcome!
Loading comments...