Zile Huang 

Welcome to my portfolio website, where I am eager to showcase my technical and business skills through a collection of exciting data analytics projects. By combining my passion for data analysis with exceptional attention to detail, I am able to turn complex insights into actionable solutions. I am actively seeking opportunities to grow as a Data Analyst and make a meaningful impact for organizations through empowering data-driven decision making.


Hi, I'm Zile Huang - a data analyst in Canada

Peter Sondergaard

Senior Vice President and Global Head of Research at Gather, Inc.

The Roll Up the Rim to Win (RUTR) campaign is a key aspect of Tim Hortons' loyalty program, and has been instrumental in improving the Canadian coffee chain's profitability. This case study analyzes 900k transactions over 9 months, with a focus on the RUTR active period of 3 months, across 50 stores in urban, suburban, and rural locations. The findings show that urban stores experience a noticeable transaction surge during and after the campaign, while rural stores do not see an increase in transactions. The higher win rate stores were found to have a higher transaction volume per customer, suggesting a rise in repeat visits. Redeeming a prize for a donut results in a higher likelihood of making an additional purchase compared to coffee. The distribution of coffee prizes is likely to result in negative profits, while donut prizes are more profitable due to their higher add-on rate and higher margin add-ons. The overall campaign was deemed to be profitable due to strong reactions in urban stores, but unprofitable in rural stores and had limited impact in suburban stores.

Get ready to delve into the world of loyalty programs and drive business success!

![](/images/WechatIMG762.jpeg)

![](/images/WechatIMG761.jpeg)

###### **(Insight 1)**

Discover the impact of the Roll Up the Rim to Win campaign on urban and rural store transactions! Urban stores experience a noticeable transaction surge during the campaign and even after it's over, providing long-lasting benefits. On the other hand, rural stores do not see an increase in transactions during or after the campaign. Get a deeper understanding of the effects of loyalty programs on different store locations.

![](/images/Picture1.jpg)

###### **(Insight 2)**

Urban and Suburban stores perform well during the RUTR campaign, while Rural stores see a daily revenue decrease during the same period.
![](/images/Picture2.jpg)

###### **(Insight 3)**

Urban and suburban stores increase staffing during the RUTR campaign, while no change in rural store staffing.

![](/images/Picture3.jpg)

###### **(Insight 4)**

If redemptions are considered to be an indicator of winning odds, then the likelihood of winning a prize is estimated to be 17%, equivalent to a 1 in 5.8 chance. It's worth noting that the distribution of prizes is slightly biased towards coffee prizes, with a +10% skew in their favor.

![](/images/WechatIMG755.jpeg)

###### **(Insight 5)**

Six stores were observed to have a significantly higher win rate compared to other stores in the same category. This was part of an experiment conducted by management to determine the impact of an increased frequency of prize redemptions on customer visits. After controlling for location, the higher win rate stores were found to have a higher transaction volume per customer, suggesting a rise in repeat visits.

![](/images/WechatIMG756.jpeg)

###### **(Insight 6)**

Redeeming a prize for a donut was found to result in a 20% higher likelihood of making an additional purchase, compared to redeeming a prize for coffee.

![](/images/WechatIMG757.jpeg)

###### **(Insight 7)**

The distribution of coffee prizes is likely to result in negative profits from incremental transactions, as the add-on rate is low. In contrast, donut prizes have a significantly higher add-on rate and higher margin add-ons, making them more profitable. It's worth noting that coffee prizes are cheaper than donut prizes due to their lower cost structure.

![](/images/WechatIMG758.jpeg)

In order to determine the optimal level of prize distribution, a market research study was conducted last month to explore the best mix of prizes for the 2019 Roll Up the Rim campaign. The study surveyed a sample of participants to gather insights and inform decision-making.

![](/images/WechatIMG743.jpeg)

###### **(Insight 8)**

**Theorizing On the Construction of P\&L Model**

**Overall Operating Profitability - Template**

![](/images/WechatIMG760.jpeg)

###### **In conclusion**

The overall campaign was deemed to be profitable due to strong reactions in urban stores. However, upon closer examination, the campaign was found to be unprofitable in rural stores and had limited impact in suburban stores.

#### **Feeding Back Insights Into Strategy Design**

![](/images/WechatIMG759.jpeg)

*   Here’s the [Tableau Viz](https://public.tableau.com/views/Tim_Hortoms/Dashboard1?:language=en-US&:display_count=n&:origin=viz_share_link) of Tim Horton' RUTR campaign performance (interactive).

![](/images/WechatIMG751-84adaf07.jpeg)


Roll Up the Rim to Win Campaign is Tim Hortons' iconic loyalty program. This case study assesses KPIs further and discover insights, which are then fed back into the business through strategy modifications.

Rolling to Profits: An Analysis of Tim Hortons' Roll Up the Rim to Win Campaign on Customer Transactions and Profitability

Fraud detection in Taxi scams

Mitigating Taxi scams through advanced data analysis: A case study in detecting third-party fraud transactions

**Project’s Questions:**

How to improve customer retention and maximize profitability?

***Executive analysis:***

**Background:**

The case study was 400k+ transactional level
data that an e-commerce company generated in China from February to April 2016.
The objectives of this project are to identify high-value customers who
purchase less with the most contribution and to develop marketing
strategies to improve customer retention and optimize profitability. The
K-means and the RFM model are the two methods I used to explore and analyze the
data in Tableau.

***The K-means model***

*   Here’s the [Tableau Viz](https://public.tableau.com/views/CustomerClusteringusingKMeans/Dashboard1?:language=en-US&:display_count=n&:origin=viz_share_link) using the K-means method (interactive).

In this case, the data was separated into 10
cluster groups.  Only four cluster groups
can be interpreted further since several clusters have substantial
similarities, and others have insufficient data, which is an obstacle to
further interpretation.

*   [Elborw method](https://en.wikipedia.org/wiki/Elbow_method_\(clustering\)#:~:text=In%20cluster%20analysis%2C%20the%20elbow,number%20of%20clusters%20to%20use.) helps to find the optimal numbers of clusters.

***The RFM model***

*   Here’s the [Tableau Viz](https://public.tableau.com/views/CustomerCentricAnalyticsusingRFMmodel/Dashboard2?:language=en-US\&publish=yes&:display_count=n&:origin=viz_share_link) using the RFM model method (interactive).

However, the RFM model can address the limitations
of the K-means since the RFM model is detail-oriented, and can combine business
knowledge. Those high-value customers, especially Lost best customers, are our
target customer groups; therefore, using the RFM model to cluster data into
five groups based on the RFM scores. The RFM scores consist of three digits: recency,
frequency, and monetary. Each rating scale is from 1 to 5. 1 refers to the
most, and 5 refers to the less. 

**Conclusion:**

The K-means model has limitations when data quantity is
limited. However, the RFM model does substantially help to cluster groups by
customer behaviors. We can combine outputs from both cluster groups in the
K-means and the RFM model for further business insights and design different
marketing strategies to improve customer retention and maximize profitability.

*   Click [Here](https://medium.com/@zile.huang.offical/data-analysis-with-tableau-customer-centric-analytics-using-k-means-and-rfm-model-a2e973231c7e) to learn more details about the project



*   The poster below indicates how we applied the marketing strategies into its website.

![](/images/IMG\_7634.JPG)![](/images/IMG\_7630.JPG)

How does an E-commence company improve its profitability and customer satisfaction? The article provides the solution using the K-means and the RFM models.

Hi, I'm Zile Huang - a data analyst in Canada

“Information is the oil of the 21st century, and analytics is the combustion engine.”

Data analytics projects

Rolling to Profits: An Analysis of Tim Hortons' Roll Up the Rim to Win Campaign on Customer Transactions and Profitability

Mitigating Taxi scams through advanced data analysis: A case study in detecting third-party fraud transactions

Improving Customer Retention and Maximizing Profitability in the E-commerce Industry: A Case Study of a Chinese Company

Zile's portfolio

Hi, I'm Zile Huang - a data analyst in Canada

“Information is the oil of the 21st century, and analytics is the combustion engine.”

Data analytics projects

Rolling to Profits: An Analysis of Tim Hortons' Roll Up the Rim to Win Campaign on Customer Transactions and Profitability

Mitigating Taxi scams through advanced data analysis: A case study in detecting third-party fraud transactions

Improving Customer Retention and Maximizing Profitability in the E-commerce Industry: A Case Study of a Chinese Company