Education

Experiences | Certifications

Skills

TECHNICAL SKILLS

PROGRAMMING :

• Python (Avanced user).
  
• Git (Avanced user)
  
• PostgreSQL(Advanced user)
  
• JAVA (Basic user)
  
• R (Intermediate user)
  
• R-Instat (Intermediate user)
  

Frameworks & Libraries:

• Pandas
  
• NumPy
  
• Matplotlib
  
• Seaborn
  
• Scikit-Learn
  
• Tensorflow
  
• Keras
  
• Jupyter

AREA OF INTEREST

• Machine Learning and Deep Learning
  
• Cybersecurity
• Computer Vision
  
• Natural Language Processing
  
• Prompt engineering
  
• Climate Dynamics
  
• Data mining
  
• Data Visualization
  
• Linear Algebra
• Calculus

Projects and case studies

Mphil/Ph.D Research Proposal submitted to University of Mauritius

Modelling Continuous and Discrete Non-Negative Integer-valued Time Series Data ​Using Machine Learning Algorithms

The Autoregressive (AR), Integer Autoregressive (INAR), and Autoregressive Integrated Moving Average (ARIMA) architectures are the most ​popular time series structures that have been successfully applied to model real-life time series problems. The inferential methods to estimate ​the model parameters in these structures include mainly the conditional maximum likelihood and least squares. These results are, therefore, ​used for forecasting. However, over recent years, the use of Artificial Neural Networks (ANN) and Support Vector Machines (SVM) to tackle the ​inferential and forecasting procedures in time series analysis is gradually capturing the attention of the researchers’ community. Such ​techniques have been mostly used in continuous time series models, including mainly the AR structures. In the same trend, Recurrent Neural ​Networks (RNN), particularly suited to model non-linear dependencies between time series, have offered promising results. A special type of ​RNN, the Long Short-Term Memory (LSTM), has been designed to improve the vanilla RNN by enforcing the long-term learning behavior when ​given the data and simulating the behavior of randomly generated data over time through complex units with varied components. This PhD ​research proposal aims to introduce and enhance innovative techniques concerning classical time series analysis and forecasting, as well as ​artificial neural networks (ANN), recurrent neural networks (RNN), support vector machines (SVM), and various statistical and machine learning ​methodologies. These methods will be applied to continuous and discrete-valued time series models for comprehensive analysis and accurate ​forecasting. The implementation of these algorithms will be followed by a comparison with different methods, such as INAR, ARIMA, and simple ​RNN, considering simulated and real-world time series data, thus making an additional and up-to-date contribution in modeling continuous and ​discrete non-negative integer-valued time series data. The proposed algorithms in this research proposal will be applied firstly on simulated ​data from the family of auto-regressive models to assess the consistency and asymptotic performance of the respective estimators and then to ​real-life time series data. The real-life time series data can be made available from the forecasting principles websites, the different R packages, ​journal archives, and other publicly available websites like Statistics Mauritius, World Health Organization, and others.

Email me: jnguinabe@aimsammi.org

WhatsApp : +237 656 74 20 77

I welcome collaboration opportunities within the ​realm of machine learning. Whether you seek to ​pioneer inventive solutions or delve into ​uncharted research avenues, count on me to ​offer valuable insights and expertise.

Feel free to reach out to me through the contact ​form or WhatsApp on this website to discuss ​potential collaborations.

I'm fluent in French, can handle ​English independently, and have ​basic proficiency in Arabic.

Thanks for stopping by my website! Let's work together to leverage the potential of machine learning ​and create a better tomorrow!