Aboubakiri DIAW

About Me

Education : I am a recent graduate from Ecole Centrale with a Data Science and Digitalization Option (S2D) - Management Track and I have two years of experience in finance. My background in mathematics and my strong analytical mindset have prepared me well for a career in data science.

Research Interests : I’m currently learning Data science that includes Machine learning and Deep learning. I’m looking to collaborate on data science projects

Publications: All of my projects are available on my github account

Projects

2023-04-10

In this project, I build a data pipeline using python. the project consists in creating a RestAPI (FastAPI) with a specific endpoint that will be able to return the Methane emissions for a country at a specific year.

1. Data retrieval: Using a Python script to retrieve methane emission data from the World Bank API using the WBGAPI module.
2. Data estimation: Using various methods such as rolling statistical, KNNImputer, and linear interpolation to estimate missing methane emission data.
3. Uncertainty computation: Using the bootstrapping method to calculate the uncertainty of the estimated values.
4. Scoring methodology: Evaluating countries based on their methane emissions using various methods such as total emissions, emissions per area, emissions per capita, and emission intensity.
5. Data visualization: Creating interactive visualizations using the power BI to present the results.
6. REST API (FastAPI) to return the Methane emissions for a country at a specific year.

Here is the link to the code

2023-03-20

In this project, we build a machine learning model to predict the score of Premier League matches. For this we have data from the last 3 seasons.

1. Data retrieval: Using a Python script to scrape premier league matches data from this website here
2. Preprocessing: Using right data types to optimize the memory used of the data and performing cleaning data.
3. Modelisation: Using machine learning model especially RandomForestClassifier. I trained the model first and predict the result of future matches.
4. First Evaluation: Compute the precision score of the prediction of the model.
5. Improving the precision: Creating new features by performing rolling average.
6. Last evaluation: Compute again the precision score of the prediction of the model improved.

Here is the link to the code

2022-09-05

The goal of this project is to find similarities with players based on their ratings. In order to achieve this goal, we implemented the k-means clustering algorithm which is an unsupervised machine learning algorithm that you can use to find clusters in your data.

1. Data cleaning: handling missing values by deleting them
2. Feature selection: Selecting 5 most important features to build clusters
3. Normalize the dataset: Scaling data between [0,4] to normalize values.
4. Modelisation: Building K-means algorithm from scratch.
5. Comparison: Comparison with the KMeans module of the sklearn library.

Here is the link to the code

2020-09-01

The goal of this project is to build and train a convolutional neural network (CNN) that will serve to classify photos of the three football players (Iniesta, Neymar, and Messi).

1. Preprocessing data: To achieve this goal, I trained the fine-tuned VGG16 model to classify images of the three football players.
2. Training: To achieve this goal, I trained the fine-tuned VGG16 model to classify images of the three football players.
3. Building a web application: I used a simple Flask App to build a web application to visualize the result of the model
4. Result: In this video here, you can see that the model predicts well the images given in input with pretty good accuracy.
5. Future plans: The next step is to build more deep learning apps using NLP, RNN models, and so on .

Here is the link to the code

Contact

Email: aboubacar198@gmail.com

Address: Casablanca-Morocco

• Linkedin
• GitHub
• Facebook
• Instagram