Machine Learning has become a must-have for all companies out there.
Your data contain a TON of hidden value you might not even realize. Machine Learning algorithms have the power to automatically harness it.
Whether you need to perform regression tasks, classification tasks, Deep Learning, Survival Analysis, or Time Series, the R language is fully equipped with the most powerful Machine Learning tools to train your models and put them into production.
With a strong background in mathematics and statistics, I master all the technical aspects of the algorithmes, from their implementation to their intepretation, without forgetting feature engineering!
I also take care of cleaning your data, enriching it with open and public data, and transform it to obtain the best performance for your business! Algorithms alone aren't sufficient!
How does it work?
Define the problem
The first step of understanding the core business issue is the most important one. We decide how to handle the problem and which KPI will be optimized with the algorithms.
Your data are extracted, cleaned, and enriched to guarantee the best performance. Variables are transformed, combined, and new features are created. Sometimes we can even use open data.
Train and optimize models
Several models are trained and evaluated to select the best one. Hyperparameters are tuned to make sure the models perform at their best. This whole process is repeated multiple< times until results are satisfying.
I can help your team in the process of setting up a Machine Learning model in production. From defining the problem, to choosing the model, or improving your results, my experience can help you get better results faster.
Build predictive models
Hand me your data and I'll take care of the project from start to finish. From extracting the data to preparing them, training models, evaluating them, optimizing them, and deploying an API to your servers.
R, caret, keras, and h2o
I am using the R language, equipped with the best libraries for Machine Learning, such as
caret for standard Machine Learning,
keras for Deep Learning, and
h2o for AutoML.
Deep Learning is an extension of Machine Learning, using very powerful algorithms used in computer vision, text mining, or complex problems that can be solved with a huge amount of data.
Put in production
Once the model is created, evaluated, and chosen, I deploy it on your servers with a REST API so that your users, your app, or your team can use it at any time and obtain predictions quickly.
Vizualisation and interpretation
Sometimes algorithms are black-box models. We can still try to visualize and interpret them (thanks to LIME), which allows you to understand what's going on under the hood. With a Shiny app, you can even experience the models yourself and make simulations to see the impacts of some features.
Work with me in 3 steps:
Tell me about your company and what you need specifically.
I write a proposal and we discuss it to add, complete and adapt any element to your needs.
While keeping you posted, I complete the project and deploy it in your company.