Freelancer profile translated to English.

Description

My mission: to help you unlock the potential of your data! Data analysis enables companies to leverage their data to achieve their goals.

Among my clients' major challenges, we identify those for which data analysis and machine learning provide a quick, effective, and sustainable solution.

My services are mainly structured around two strategic axes:

1 - Optimization of your operations with data analysis

Churn reduction
Cost reduction
Revenue increase
Quality improvement

2 - Anticipation of your activities with predictive models

Production prediction
Demand forecasting
Unpaid prediction
Failure and breakdown prediction

Client references: Sanofi, Keolis, Kinkelder, Crédit Agricole, Touton SA...

Languages

French
Native or bilingual
English
Native or bilingual

Workplace preferences

Can work on-site

Bordeaux (up to 50km), Paris (up to 50km), Toulouse (up to 50km), Lyon (up to 50km), Lille (up to 50km)

Keolis
Ticket Sales Prediction at Keolis
TRANSPORTATION
July 2020 - September 2020 (3 months)
Bordeaux, France
CONTEXT

The marketing department of Keolis wishes to leverage their data to predict ticket sales from vending machines.

The two main objectives of the mission are revenue prediction and network maintenance improvement.

Objectives:

*Predict ticket sales

*Identify useful data

Approach:
Data consolidation
Predictive Machine Learning

Benefits:
Anticipate revenue
Anticipate activity flows

APPROACH

1 - Correction of missing data
2 - Creation of replacement variables
3 - Use of kernel algorithms

RESULTS OBTAINED

The developed Machine Learning tool allows for sales prediction with an average accuracy of 96%.

Summary of the 12-week project:

Custom predictive tool integrable internally
Detailed documentation and data mapping
Training days and educational materials
Machine learning Data science Big Data Data Engineer Data analysis
Touton
Machine Learning Model for Cocoa Production Prediction
IMPORT AND EXPORT
March 2020 - June 2020 (3 months)
Bordeaux, France
Objectives:

*Make production predictions more reliable

*Verify input data

Approach:
Data exploration
Algorithm validation and Machine Learning

Benefits:
Anticipate production
Better market positioning

Context:
To best develop its new activities, Touton provided historical data to work on improving the statistical model used to forecast Cocoa production in their plantations in Ivory Coast and Ghana.

My mission is to analyze all available data, classify it, determine its relevance, and understand how it is generated, in order to build the best possible predictive algorithm.

Methodology:

1 - Correction of missing data
2 - Creation of new variables
3 - Use of kernel algorithms

Results:

The work done has reduced the error rate of the cocoa production prediction tool by 20%.

The tool we developed surpassed the performance of the existing model after only 16 weeks of work. Predictions are now more accurate, there are fewer errors in the data types, and the restructuring of variables allows for continuous improvement of the tool, ensuring Touton an incremental ROI on the project.
Data science Big Data Machine learning Database administration Data Engineer Data analysis
Sanofi
Logistics Cost Reduction
MEDICAL
January 2020 - March 2020 (1 month)
Bordeaux, France
Eager to leverage the data generated by its operations, a Sanofi distribution site entrusted me with 18 months of historical data to work on optimizing logistics costs, a central issue for a company that transports thousands of products daily.

We developed our thinking around two axes:

What data is used?
What meaning should be attributed to it in the studied context?

To address the issue of transport cost optimization, our team worked simultaneously on the different facets of the problem. Algorithms specific to the identified fields of action were applied to independent datasets to validate the chosen model for learning.

Results: €200,000 in savings at a distribution site

ACTIONS ON SETTINGS:

The search for the origin of anomalies generated by the information system allowed us to identify the variables responsible for the situation and to make corrections that yielded rapid results.

ACTIONS ON PROCESSES:

For some of the highlighted issues, modifying processes, such as order handling or stock management based on volume forecasts, provides applicable solutions.
Data science Big Data Machine learning Data visualization