🧪 A/B Testing for Conversion Optimization

A real-world simulation of an A/B test used by product teams to make data-driven decisions on conversion optimization.

---

📌 Problem

A product team aims to improve landing page conversions. Two versions are tested:

• Variant A (current version)
• Variant B (new version)

The goal is to determine whether the new design leads to higher user conversions.

---

🎯 Objective

Analyze user conversion data and determine whether Variant B significantly outperforms Variant A using statistical testing.

---

📊 Dataset

• 8,000 simulated users

Features:

• user_id → Unique identifier
• group → A / B test group
• converted → (0 = No, 1 = Yes)

---

⚙️ Approach

• Cleaned and validated dataset
• Calculated conversion rates for both groups
• Performed Z-test for proportions

Hypothesis:

• H₀: No difference in conversion rates
• H₁: Variant B improves conversions

---

📈 Results

Conversion Rate:

• Variant A: 13.0%
• Variant B: 12.6%

Statistical Test:

• Z-statistic: 0.5225
• P-value: 0.6013

👉 Conclusion: No statistically significant difference between the two groups

---

💡 Insight

Variant B does not improve conversions and should not be implemented based on statistical evidence.
The observed difference is due to random variation, not a meaningful change in user behavior.

---

🚀 Recommendation

• ❌ Do not implement Variant B
• ✅ Retain current version (Variant A)
• 🔄 Test more impactful changes in future experiments

---

📊 Visualization

---

🛠️ Tech Stack

• Python
• Pandas
• NumPy
• Statsmodels
• Matplotlib

---

▶️ Run the Project

git clone https://github.com/Harshu2326/AB_Test_Project.git
cd AB_Test_Project
pip install -r requirements.txt
python main.py