top of page

Analytics Projects

The projects below are some examples of my work during my Masters's in Business Analytics. I had the opportunity of working with real data sets such as Whole Foods, Air France, and Wells Fargo. During this program I learned how to manage big data sets, import them to SQL, Python, or R, perform analysis, identify insights, and present them using data visualization tools such as Tableau. 

​

This experience not only enriched my marketing knowledge but also equipped me with hard skills, practical experience and confidence in tackling complex business analytics challenges. 

To see more projects I invite you to check out my github page: https://github.com/karlalobo/projects

Air France - Campaign performance analysis using R​

​

Our goal with this analysis was to optimize Air France’s marketing budget while increasing sales. We began our analysis by creating a marketing score with different weights depending on the importance of the variables, we chose Transaction Converted Percentage to have the highest weight of 40%, followed by Click Through Rate with 30% of the weight, the Cost Per Click and the Total Cost with 15%.

 

With this formula, we compared the 7 publishers and we identified that the top 3 are Google Global, Google Us, and Yahoo US. Each one of these publishers had its most successful campaign, therefore, we recommended tailoring a strategy for each publisher to maximize the return on investment. 

 

For example, the campaigns that would be the most successful in Google Global are:

  • Increase Cost Per Click: by concentrating on high-performing ad placements that will perform better

  • Improve the Ad Rank by boosting the keyword quality and enhancing the landing page experience

  • Targeted Ad Campaigns by targeting the customers by location, demographics, and interests to guarantee that our ads are being viewed by our target audience.

  • Focus on brand Awareness by creating a strong social media presence that emphasizes our unique selling propositions

 

To maximize the ROI we need to increase the CPC as the market is really saturated and very competitive. Having a higher CPC we can reach our target audience better, and we remain competitive with high visibility and improve our positioning of our ads on SEO.

 

To finalize the analysis, we evaluate the odds of success using the formula in R where we identified that, for example, in google global for every increase in the cost per click, the odds of business success would increase by 21%.

Screen Shot 2023-07-19 at 22.16.52.png
Screen Shot 2023-07-19 at 22.17.05.png
Screen Shot 2023-07-19 at 22.17.26.png

Whole Foods- Performing an analysis in SQL to answer: Do dietary preferences affect price?​

​

According to the t-test performed with the price and the total score of dietary preferences, the P value is less than 0.5, therefore it confirms that there is a significant relationship between dietary preference and price (appx 1). However, by comparing these results with the regression analysis and the comparative table (appx 2 & 3), the results are the opposite and they show no significant relationship between the variables.

 

Nonetheless, there is enough research that confirms the relationship between dietary preferences with higher costs. According to Koebert (2022) “Since 2019, the average cost for foods essential to popular diets has risen by 11.64% on average. The overall inflation rate in that same time frame is just 9.05%”. According to Harvard (2013), The healthiest diets cost about $1.50 more per day than the least healthy diets; and according to (Drewnowski, 2005), added sugars and added fats are far more affordable than the recommended “healthful” diets based on lean meats, whole grains, and fresh vegetables and fruit.

 

Therefore, we can conclude that dietary preferences do effect the price. The reason why we are getting contradictory results is because each category is very different, and it is very difficult to compare one to the other, as many other variables come into play.

​

Recommendations to Whole Foods

​

Actionable insight #1: Direct efforts in marketing towards Keto-friendly. This will not only help to reach more potential customers but maintain their brand positioning, as a high-quality supermarket and increase revenue.

  • Based on the comparative table (appx 3), we can identify that Keto is the Dietary Preference that on average has the highest price.

​

Actionable insight #2: Invest in SEO keywords related to dairy-free products to increase traffic to the website through this popular dietary preference. Furthermore, invest in research on how actual dairy-free products are selling and if there is a demand for a more variety of dairy-free / dairy alternative products.

  • Based on the comparative table (appx 3), we can also identify that dairy-free is the Dietary Preference with the greatest number of products. Additionally, according to Children’s Hospital (2022), 30 to 50 million Americans are lactose intolerant.

​

Actionable insight #3: Invest in data collection about how the dietary preferences are performing. Identify the ones that are selling the most, the least, and the one that attracts the most customers to Whole Foods.

Regression Model Development in Python

​

This assignment focuses on analyzing Apprentice Chef's customer database to identify the variables that have the most significant impact on the business's revenue. The database contains approximately 2,000 customers who made specific purchase patterns during their first year. The objective is to optimize sales efforts and increase revenue through data analysis and model development.

​

The study begins by examining customer data, cleaning it, and creating new features to improve data interpretability and model performance. Several key observations are made, including data types, missing values, data skewness, and the need for data cleaning before analysis.

​

To gain better insights into customer behavior, five new features are created, such as "AVG_NUM_MEALS" and "ENGAGEMENT," which provide valuable information for revenue improvement. Three regression models (Decision Tree Regressor, Random Forest Regressor, and Gradient Boosting Regressor) are constructed and evaluated, with the Random Forest Regressor performing the best, indicating its suitability for the dataset.

​

The results highlight the importance of thorough data analysis and feature engineering in developing accurate predictive models. The selected model, Random Forest Regressor, demonstrates good predictive power, with 84% of the features showing strong correlation to revenue. The findings emphasize the significance of conducting a comprehensive analysis for accurate and reliable results in revenue optimization.

bottom of page