Listen to this articleBelow is our recent interview with Simon Gazikian, CEO of MyDataModels:
Q: Could you provide our readers with a brief introduction to MyDataModels?
Simon Gazikian: We are a software publisher, incorporated in March 2018 in Sophia-Antipolis, France, after more than 5 years of R&D. MyDataModels’s vision is to give Machine learning access to everyone, we mean every professional in companies. And because in companies, more than 80% of data are Small Data, our software empowers domain experts to extract predictive value from their Small Data. Domain experts – engineers, researchers, analysts, marketers, and consultants with relevant data – can generate and run predictive models on their own data in a simple and automated way, without knowledge of programming and Machine Learning. They don’t need support of data scientists, they are are fully autonomous.
Recommended: ZenythGroup Offers A Deep Expertise In ADA Website Accessibility And WCAG Compliance For Web And Mobile Applications
Q: Can you tell us something more about TADA? How does it work?
Simon Gazikian: TADA is easy to use, it does not require coding nor machine learning skills. Inspired by evolutionary algorithms, TADA is particularly suited for Small Data which are actually departmental data that domain experts deal with on a daily basis. We have removed major roadblocks of this category of algorithms to make them usable in companies. Its main benefits are:
1. Empowerment of domain experts, self-service predictive modeling.
2. Quick wins: models are built and deployed in few days versus weeks/months.
3. Unique ability to create predictive models from small samples of historical data, Small Data
4. Web-based intuitive interface (including data visualization and decision tools),
5. Ability to quickly identify the factors having the most impact in a given situation.
Q: What are the main features of your platform?
Simon Gazikian: 1. Ease of use: Web-based intuitive interface, one-click generation model, data visualization and decision tools, video tutorials at every single step of the process,
2. Data preparation is strongly reduced: automated preprocessing: missing values, stratified sampling, one hot encoding. There is no need to do feature engineering, manage outliers or align values range. TADA produces better results with raw unaltered data as “noise” in a dataset can itself contain predictive meaning.
3. Automated variable reduction: ranking of most significant variables impacting target variable, capacity to reduce several thousands of variables down to very few, consideration of 50k variables is very practical from performance perspective even with modest hardware.
4. Modeling process: automatically creates training, validation and test partitions, no need to tune hyperparameters , creates compact predictive models made of 3 to 7 variables max, generates a predictive model in a few minutes, runs in cloud (public/private), desktop (PC, Mac, Linux), mobile and edge computing environments.
5. Modeling capabilities, model review and deployment.
Recommended: The SSL Store Provides A Variety Of Digital Security Solutions For SMBs And Large Enterprises
Q: WWhat are some of the specific use cases or scenarios you can share on how your clients can leverage TADA?
Simon Gazikian: TADA is not limited to specific use cases as long as they are based on numeric and structured data. We have plenty of use cases scenarios already available: Quality prediction (Operations), Heart Disease prevention (Research), Marketing Campaign performance prediction, Backorder prediction (Supply chain), Breast cancer prediction (Research), Churn Prediction (Customer services), Fraud Detection (Security), Supply chain optimization (Manufacturing), failure prediction (manufacturing), product development (Research and development)
Q: What are your plans for the future?
Simon Gazikian: Become the leading digital company providing machine learning solutions for everyone. We are strongly growing our customers installed base and our network of partners – in technology and IT services.
Activate Social Media:
