Panthéon-Sorbonne

Major in Data Science

I was enrolled in a diverse range of courses at Panthéon-Sorbonne University, which include valuable theoretical and practical components, as well as engaging in several Semestrale projects, that allow me to apply the knowledge and skills acquired in my classes to real-world scenarios, and work on research-oriented initiatives.

In today's competitive business environment, retaining valuable customers and ensuring the wellbeing of employees is paramount. My project, "Leveraging AI and Machine Learning for Proactive Churn Prediction and Employee Burnout Detection", showcases the potent power of predictive analytics and machine learning in meeting these challenges.

I utilized historical customer data and applied advanced machine learning models (like Random Forests, Logistic Regression, etc.) to proactively predict customer churn. This predictive methodology grants companies the critical time to strategize and implement retention measures, thereby boosting customer loyalty and increasing revenue.

On the employee front, I employed innovative AI algorithms to analyze employee performance data, spotting early signs of burnout. These predictive insights allow organizations to intervene at an early stage, promoting a healthy, motivated, and productive workforce. This strategy enhances job satisfaction, curbs staff turnover, and dramatically reduces costs associated with employee burnout.

This project demonstrates how data-driven decision-making can lead to more strategic and effective business operations. A perfect blend of data science and business acumen, it highlights my capabilities in using AI and machine learning to tackle real-world business challenges. As a prospective employee, I am excited to bring these skills to the forefront.

In the rapidly evolving landscape of technology, understanding consumer sentiment and expectations is key. My project, "Consumer Sentiment Analysis: Harnessing the Power of AI to Understand User Reviews", illuminates the strategic power of data analysis and machine learning in deriving valuable insights from customer reviews.

For this project, I sourced a dataset of iPhone 13 consumer reviews from various French retail stores. To ensure a representative and unbiased sample, I included both positive and negative comments. The reviews were then subjected to a rigorous analysis process using the R programming language and the RStudio IDE.

I used the 'tm' library for text processing and corpus creation. Data preparation steps involved cleaning the comments (lower-casing, removing numbers and unnecessary words, replacing punctuation with space, and eliminating extra spaces), and creating a corpus using R's 'corpus' function.

To visually represent the most frequently occurring words in the reviews, I employed the 'Wordcloud' library.

For deeper analysis, I used Correspondence Analysis (CA) with the help of the 'FactoMineR' and 'FactoInvestigate' libraries. This allowed me to identify the proximity of consumer profiles to certain terms and highlight the significant factors in customer feedback.

My CA visualizations indicated a strong correlation between product quality and customer feedback. I discovered that elements like photo quality held great importance for customers seeking quality products. I was also able to identify and group customers based on their experiences, forming clusters that could be invaluable for targeted marketing strategies.

Finally, Ascending Hierarchical Classification was employed to reveal six distinct clusters of customers, each characterized by different feedback elements.

This project demonstrates the potential of data analytics and AI in offering a sophisticated understanding of consumer sentiment. It highlights my ability to harness these technologies to provide actionable business insights. As a prospective employee, I am eager to apply these skills to new challenges.

This master's thesis in Data Science was a group project that used statistical analysis and machine learning techniques to examine the correlation between gambling and various other factors, such as alcohol consumption, cigarette use, gender, and region. The study found that there is a positive correlation between gambling and these factors, with the strongest correlation being between gambling and alcohol consumption. Predictive models were also used to predict whether an individual would gamble based on their responses to other variables, and the most important variables for the model were found to be alcohol, cigarettes, and region. The study also used decision tree and random forest methods and concluded that the minimal error rate in the random forest model was reached with 10,000 trees. Overall, this study provides insights into the relationship between gambling and other factors using data science techniques.

This group project in Data Science aimed to investigate the relationship between dishonesty and behavior towards gambling. Five hypotheses were formulated and tested, but none of them were fully or even partially validated. However, the study did identify a trend in the way dishonest individuals tend to play, they tend to prefer a specific type of game and generally play less aggressively, preferring to make low-risk bets with small sums. The study also noted some limitations, such as the small sample size (156 respondents, with over 70% being female) which may not be representative of the population, the subjectivity of measuring dishonesty using only six variables and the social desirability bias that can affect the objectivity of the participants in their responses to the questionnaire.