https://github.com/mnitin-reddy/a-b-testing-and-regression-analysis-for-ad-performance-optimization
Analyzed the performance of Facebook and AdWords ads using A/B testing and regression analysis to identify trends, correlations, and cost-effectiveness. Key insights included distribution of clicks and conversions, monthly trends, and cost-per-conversion analysis to optimize ROI.
https://github.com/mnitin-reddy/a-b-testing-and-regression-analysis-for-ad-performance-optimization
abtesting data-science hypothesis-testing machine-learning matplotlib numpy pandas scikit-learn scipy seaborn statsmodels
Last synced: about 1 month ago
JSON representation
Analyzed the performance of Facebook and AdWords ads using A/B testing and regression analysis to identify trends, correlations, and cost-effectiveness. Key insights included distribution of clicks and conversions, monthly trends, and cost-per-conversion analysis to optimize ROI.
- Host: GitHub
- URL: https://github.com/mnitin-reddy/a-b-testing-and-regression-analysis-for-ad-performance-optimization
- Owner: MNitin-Reddy
- Created: 2024-12-13T09:05:26.000Z (5 months ago)
- Default Branch: main
- Last Pushed: 2024-12-17T06:02:59.000Z (5 months ago)
- Last Synced: 2025-02-08T11:44:02.477Z (3 months ago)
- Topics: abtesting, data-science, hypothesis-testing, machine-learning, matplotlib, numpy, pandas, scikit-learn, scipy, seaborn, statsmodels
- Language: Jupyter Notebook
- Homepage:
- Size: 1.33 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# A/B Testing and Regression Analysis for Ad Performance Optimization 📊
## Business Problem
As a marketing agency, our objective is to **maximize the return on investment (ROI)** for our clients' advertising campaigns. We conducted two ad campaigns:
1. **Facebook Ads**
2. **AdWords Ads**
We aim to determine which platform performs better in terms of:
- Clicks
- Conversions
- Cost-effectiveness
The insights will help optimize resource allocation and enhance advertising strategies.
---
## Research Question
**Which ad platform is more effective in terms of conversions, clicks, and overall cost-effectiveness?**
---
## Dataset Description
The dataset contains daily performance metrics for both Facebook and AdWords campaigns throughout **2019**. Key features include:
- **Date**: Campaign date (2019-01-01 to 2019-12-31)
- **Ad Views**: Number of views on the ads
- **Ad Clicks**: Number of clicks received on the ads
- **Ad Conversions**: Number of conversions resulting from the ads
- **Cost per Ad**: Cost associated with running the ad campaigns
- **Click-Through Rate (CTR)**
- **Conversion Rate**
- **Cost per Click (CPC)**
---
## Technologies Used
- **Python Libraries**: pandas, numpy, matplotlib, seaborn, scipy, statsmodels, scikit-learn
- **Tools**: Jupyter Notebook
- **Statistical Techniques**: Correlation Analysis, Hypothesis Testing, Regression Modeling, Cointegration Test
---
## Analysis and Insights
### 1. Distribution of Clicks and Conversions
The distributions of **clicks** and **conversions** show a symmetrical shape, indicating even performance across campaigns.
**Visualization:**
---
### 2. Frequency of Conversions by Categories
We categorized conversions into:
- Less than 6
- 6–10
- 10–15
- More than 15
**Observations:**
- Facebook had more frequent high-conversion days.
- AdWords lacked days with conversions above 10.
**Visualization:**
---
### 3. Do Clicks Lead to More Conversions?
We analyzed the **correlation** between clicks and conversions for both platforms:
- Facebook: **Strong Positive Correlation** (0.87)
- AdWords: **Moderate Positive Correlation** (0.45)
This suggests Facebook is more effective in driving conversions.
**Visualization:**
---
### 4. Monthly Conversions Over Time
Conversions were analyzed across months:
- Mondays and Tuesdays had the highest conversions.
- Monthly trends showed **increased conversions** over time with minor dips in February, May, and November.
**Visualization:**
---
### 5. Monthly Cost Per Conversion (CPC)
We analyzed the **Cost Per Conversion** to understand advertising cost-effectiveness:
- **May and November** had the lowest CPC values.
- February had the **highest CPC**, indicating less cost-effectiveness.
**Visualization:**
---
### 6. Hypothesis Testing
- **Hypothesis**: Facebook has more conversions than AdWords.
- **Result**:
- Mean Conversions:
- Facebook: **11.74**
- AdWords: **5.98**
- **p-value**: Extremely small (9.35e-134) → Reject the null hypothesis.
**Conclusion**: Facebook ads generate **significantly more conversions** than AdWords.
---
### 7. Linear Regression: Predicting Conversions from Clicks
We built a Linear Regression model for Facebook ads:
- **R² Score**: 76.35%
- **Insights**: Predicting Facebook conversions based on clicks helps set realistic goals.
Example predictions:
- **For 50 clicks → ~9 conversions**
- **For 80 clicks → ~14 conversions**
**Visualization:**
---
### 8. Cointegration Test: Cost and Conversions
A **cointegration test** showed a long-term equilibrium relationship between ad cost and conversions, suggesting stable budget impacts over time.
---
## Recommendations
1. Allocate **more resources to Facebook Ads** due to their higher conversions and stronger ROI.
2. Optimize AdWords campaigns to improve click-to-conversion performance.
3. Increase ad spend during months with **lower CPC** (e.g., May, November).
4. Monitor and analyze performance during **Mondays and Tuesdays** for targeted campaigns.
---
## Conclusion
The analysis demonstrates that **Facebook Ads outperform AdWords** in driving conversions. Businesses can leverage these findings to improve ad performance and ROI.
---
## How to Run 📥
1. Clone the repository:
```bash
git clone https://github.com/yourusername/ad-campaign-analysis.git
cd ad-campaign-analysis
```
2. Install dependencies:
```bash
pip install -r requirements.txt
```
3. Run the notebook:
```bash
jupyter notebook notebooks/ad_campaign_analysis.ipynb
```