In order to meet your different needs in terms of data analysis, Ondalys offers a wide range of services.
Our core business focuses on the development of spectroscopic calibrations, the optimization and updating of calibration models (for quantitative regression, or classification), the improvement of their robustness and inter-instrument transfer, the fusion of multi-block data, the supervision of processes (MSPC, BSPC), the analysis of hyperspectral images or the implementation of experimental designs.
You have developed your own models and wish to have them validated by experts in Chemometrics and Machine Learning? Ondalys offers you an audit formula for your models and procedures, in order to provide you with recommendations to optimize your work.
Ondalys advises you in the choice of your spectrometer, in the optimization of the data acquisition chain or on the constitution of the databases thanks to the experimental designs.
Proof of Concept / PoC
Thanks to the multidisciplinary skills in data analysis and Machine Learning (ML), but also thanks to the know-how in the interpretation of instrumental data and various application sectors, Ondalys team carries out feasibility studies from your data whatever they are: spectroscopic, imaging, physico-chemical analyzes, process parameters,…
Ondalys carries out its studies using ML methods, but demystifies these “black boxes” by interpreting their results thanks to sensor and trade knowledge and therefore by validating the models with human expertise!
Model development and validation
Ondalys develops prediction models that meet your needs, whatever they may be: calibration, prediction, identification, discrimination, clustering, supervision, etc.
Our approach is based on the most suitable algorithms and Machine Learning methods, to develop precise, robust and validated models, so that they can be used routinely in laboratory or industrial conditions.
Ondalys is capable of implementing your calibrations on the majority of data analysis software on the market (Unscrambler ®, SOLO® or PLS_Toolbox ® in MATLAB ® environement, SIMCA ®, etc …), including those supplied by spectroscopy equipment manufacturers (OPUS®, WinISI® , NIRCAL®, TQ Analyst®, etc.) as well as open source software (Python ®, R).
Ondalys adapts to your Chemometrics or Machine Learning software and delivers compatible models to you.
Model update and maintenance
Once implemented, Ondalys ensures the updating and maintenance of your models when necessary. Throughout the life of a model, various events can lead to an update of the model: addition of new samples, evolution of processes or raw materials, maintenance of sensors or correction of drifts, variation of the conditions of measurements, etc…
Ondalys team offers training in data analysis and Machine Learning.
These training courses, from beginners to advanced, cover a wide range of Chemometrics and Machine Learning methods: from basic methods such as PCA – Principal Component Analysis – and PLS – Partial Least Squares Regression – to methods more advanced such as SVM – Support Vector Machines -, ANN – Artificial Neural Networks – or decision trees (TreeNet/boosting, CART, Random Forests), not to mention methods of experimental designs and multi-block data fusion.
To go further, Ondalys offers you a coaching formula which allows specific support according to your level of knowledge and your issues. We work hand in hand to help you process your data in the most optimal way possible.
Our expertise in data analysis
With more than 15 years of experience in data analysis (Chemometrics and Machine Learning), in particular applied to spectroscopic, analytical and process measurements, the experts of our teams accompany you at each stage of your projects.