Our Services

Analyse icon

Statistical analysis

Machine icon

Predictive modeling

IA icon

Machine Learning

Our work process

Expression of need

This step is crucial.
Before defining the technical solution, let’s take the time to identify your final need: Reduce costs? Help decision-making? Automate a task?
As well as your technical and temporal constraints.

Data collection and exploration

Data is our raw material. Is it accessible, structured, voluminous? What defects does it present? Heterogeneous, incomplete, presence of outliers or erroneous values​​… Our data visualization will allow us to see more clearly.

Model development

There is no single model to meet all needs. Among the arsenal of machine learning techniques that we master, we will identify the most appropriate. The most important is an infallible performance evaluation and comparison methodology.

Application/API

Once the model is validated, it is time to deliver its results in a form usable and operable by your end users. The form of the deliverable is once again determined with you, according to your needs.

Machine Learning use cases

Credit Scoring

Preacor.fr

Scorelab develops for Ashler & Manson the solution Preacor.fr.

Based on Ashler & Manson brokers’ experience, combined with the machine learning algorithm developed by our team, PREACOR is enriched with new applications added daily.

Thousands of anonymized loan applications are used for the training and development of the PREACOR model. Dozens of criteria have been added to the score calculation to make the PREACOR model robust and innovative.

Preacor
Recommender system

Liv-ex

Liv-ex is the leading international marketplace for fine wines. We developed a personalized recommendation engine of wine lots for the buyers of the platform.

We built a hybrid system that uses both transaction history and product characteristics (price, appellation, vintage, etc.). Our system uses an advanced factorization model. He learns the hidden profile of every trader and every product. The recommendation lists take into account the degree of matching between customer / product profiles as well as the novelty.

Liv ex
Clustering

Genomics meta-analysis

Scorelab collaborates with IECB (European Institute of Chemistry and Biology) on a genomic study. This is a broad meta-analysis of multiple genomic experiments on the nematode Caenorhabditis elegans. This analysis brings together 1600 series of experiments using RNA sequencing technologies. The data used come from studies published on NCBI (US National Center for Biotechnology).

The purpose of this work is to understand the functions of genes and the types of cells in which the genes are active. Our statistical tools make it possible to compare the experiments with different degrees of sensitivity. This work aims to reveal unexpected correlations between genes or experiments.

Genomic capture