Projects

Forecasting

Thousands of bikes are rented every day in major cities like London. With variables such as Temperature and Holidays affecting user behavior, can we make accurate demand predictions several months ahead ?

Product Recomendation

Compare traditionnal Market Basket Analysis techniques and modern Machine Learning tools in their ability to make most relevant product suggestions.

Artificial neural network in real time
(available online)

Click here to compete against an artificial neural network integrated in a web / mobile app.

The AI model is adapted to the user's skill level by variable perturbation of its parameters.

WebApp: Name Visualization
(visualization & code available,
generation in development)

Temporal Convolutional Neural Networks and Transformers are particularly suited to efficient processing of sequential data. Here we will use such models to generate french names that fit a certain gender, time period and region specified by the user.

- Click here to visualize the data that will be used to train it.

- Click here to view the python webapp code.

Dashboard

A synthetic presentation of relevant indicators allows effective decisions to be made. Click here to access such an example of a dashboard.

Data Analysis & Visualization
(coming soon)

Visualization and Statistical testing guide us in interpreting user response to social media ads based on age and gender.

Probability Theory

Bayesian Analysis and Inverse CDF Sampling are useful tools for deriving probabilistic models.

Employee Schedule Optimization
(coming soon)

Linear and Non-Linear solvers can help us optimize a variety of business objectives. Here we automatically solve the problem of employee shift schedule based on requirements like total daily hours and number of employees needed.

Data Collection,
Webscrapping
(code available)

Large amounts of data allow us to derive interesting insights and useful models. Thankfully, we can automate and speed up the process of gathering years of information, with a process called Webscrapping.

Click here to access the code.

Data Visualization

Clear presentation of relevant information helps better formulating, understanding and achieving data-driven objectives.

Pre-Processing
(code available)

Pre-processing is an important step to ensure our model can extract as much value as possible from the data we feed it.

Click here to see an example of data pre-processing, where we check for correlation and multicolinearity between numerical and categorical variables using different types of correlation functions, in order to select the most predicitve variables, and avoid the most redundant ones.

Logistic Regression
(code available)

Logistic Regression, which forms the basic unit of Artificial Neural Networks, can be a useful algorithm on its own.

Click here to access an example of Regularized Polynomial Logistic Regression coded with Numpy, a basic Python library for scientific computing.

Autonomous Car Driving
(available online,
in development)

Use Supervised / Reinforcement Learning techniques to teach a car to navigate its way through a dynamic setting.

For now the cars have been programmed "traditionnally" using logic rules, but these will later serve as the environment to train deep learning algorithms.

Data Analyis Anchor

Schedule Opt Anchor

Projects

Data Analysis & Visualization (coming soon)

Employee Schedule Optimization (coming soon)

Data Analysis & Visualization
(coming soon)

Employee Schedule Optimization
(coming soon)