Dr. Ari Zelmanow’s Post

Check out this informative guide our incredible MLE Melanie Chen put together titled “Feature Store at Work: A Tutorial on Fraud and Risk”. In this piece, she deep dives into how ML and data teams can leverage a real-time feature engine like Chalk to build out fraud and risk models. Link to the article in the comments!

Our very definition of "data" may be too narrow. When we think of data, we picture electronic records and numbers in a spreadsheet. But human memory, passed down wisdom, and lived experience are all forms of data too. This encoded "wetware" in our minds, while not always the most reliable, can be more relevant and valuable than even the cleanest computer data set. Trusting our intuition, which is built on a lifetime of learning, can lead to better outcomes than over-optimizing based on limited external data. Every choice we make is data-driven - the key question is which data we choose to prioritize. Sometimes, the most important insights are already inside us, not on a server. — What role does gut instinct play in your most important decisions? Hear more from Cassie Kozyrkov on supplementing data with human judgment in the latest episode of The Augmented Life, out now at https://lnkd.in/gNhjqiyi

Our very definition of "data" may be too narrow. When we think of data, we picture electronic records and numbers in a spreadsheet. But human memory, passed down wisdom, and lived experience are all forms of data too. This encoded "wetware" in our minds, while not always the most reliable, can be more relevant and valuable than even the cleanest computer data set. Trusting our intuition, which is built on a lifetime of learning, can lead to better outcomes than over-optimizing based on limited external data. Every choice we make is data-driven - the key question is which data we choose to prioritize. Sometimes, the most important insights are already inside us, not on a server. — What role does gut instinct play in your most important decisions? Hear more from Cassie Kozyrkov on supplementing data with human judgment in the latest episode of The Augmented Life, out now at https://lnkd.in/gKS9bgKV

When I run one of my "Thinking Like a Data Scientist" workshops/classes, I demand that the workshop include business and operational stakeholders from the frontlines of the organization. Demand! Why? One of the keys to successful data science and AI projects is identifying variables and metrics that are better predictors of behaviors and performance. It's that accumulated "tribal knowledge" collected and refined at the frontlines of customer engagement and operational execution through years of irreplaceable experience. As this post so importantly explains: "This encoded "wetware" in our minds, while not always the most reliable, can be more relevant and valuable than even the cleanest computer data set. Trusting our intuition, which is built on a lifetime of learning, can lead to better outcomes than over-optimizing based on limited external data." Thanks for the pointer, Jean-Philippe Martin! #DataStrategist #DataScience #IOT #BigData #AI #ML #DataTransformation #DataManagement #DataEconomics #DesignThinking #GenAI #AILiteracy #DataLiteracy #IWork4Dell #AI4IA Dell Technologies Iowa State University - Ivy College of Business Coe College Menlo College University of San Francisco School of Management Mouwafac Sidaoui, VP Academic Affairs and Dean School of Business NUI Galway Business Analytics Society Denis Dennehy Amy Porter Renée B. Lahti John Morley

Working in machine learning within the industry really opens your eyes: many challenges aren't flashy, but they're just as important. Along with my colleagues Christian Ramírez Arévalo, Moacir Antonelli Ponti, and Luis Argerich, we've written a paper on one of these key challenges: how to separate the machine learning layer from business logic. This approach not only improves system robustness and scalability but also has a real impact on operational efficiency. Our study shows how this integration can lead to better business outcomes. I invite you to read our work and share your thoughts. Your experiences can further enrich this discussion! #mercadolibre #machinelearning #machinelearningengineer

Biases inevitable. I'd say, no matter how mindful. AI (risk or otherwise) modeling through data will always suffer from biases. Since in neither the model nor the data set can we humanly identify all the biases, let alone avoid them - inadvertent assumptions, incorrect norms, overlooked exceptions, etc. across so many dimensions! If trained heavily enough on pre-Renaissance (Galileo Galilei) literature, the AI model will even claim the sun still revolves around the earth! (and not the other way around) Instead, we will need a new approach on a different level... to offer a complete, objective picture along with transparent explanation of such.

Our very definition of "data" may be too narrow. When we think of data, we picture electronic records and numbers in a spreadsheet. But human memory, passed down wisdom, and lived experience are all forms of data too. This encoded "wetware" in our minds, while not always the most reliable, can be more relevant and valuable than even the cleanest computer data set. Trusting our intuition, which is built on a lifetime of learning, can lead to better outcomes than over-optimizing based on limited external data. Every choice we make is data-driven - the key question is which data we choose to prioritize. Sometimes, the most important insights are already inside us, not on a server. — What role does gut instinct play in your most important decisions? Hear more from Cassie Kozyrkov on supplementing data with human judgment in the latest episode of The Augmented Life, out now at https://lnkd.in/dC-X44rz

Data Nerds, what entertainment media do you love from an analytical perspective? 🤓 I'll start: Deal or No Deal Island. Now, DOND was always a fascinating watch from a game theory perspective. Just watching the chance elements of the case picking process, the mental calculus of the contestants, the Banker's offers and how they were used to further the drama of the show... plenty for a math geek to chew on. But this new version takes it up a notch, implementing interpersonal strategy elements from another analytical favorite of mine, Survivor. So tell me, what's your favorite show/movie/other media you love to analyze (over-analyze)? 🪖

The Price is Right, especially with the pricing rounds

1st Project : 🌟 Bankruptcy Prevention Report 🌟 Excited to share my latest project on predicting and preventing bankruptcy! This comprehensive report features high-accuracy models, including Random Forest, Decision Tree, SVM, KNN, and Naive Bayes, all demonstrating impressive performance in predicting financial distress. 📊 Key Insights and Highlights: Achieved 100% accuracy with Random Forest and Decision Tree classifiers. Detailed analysis of critical financial risk factors such as industrial risk, management risk, financial flexibility, credibility, competitiveness, and operating risk. Clear visualizations including histograms, boxplots, and distribution charts for better data interpretation and decision-making. Efficiently differentiating between bankruptcy and non-bankruptcy cases using advanced machine learning models. In-depth evaluation of financial health indicators and strategies for predicting and preventing bankruptcy. 🔍 Visual Data Representations: Comprehensive histograms and boxplots that provide clear insights into key financial risk factors. Distribution charts of scaled features and target variables for detailed understanding. Visual tools that aid in presenting actionable insights for financial analysts and decision-makers. This report is a significant resource for improving financial stability and risk management. It offers a strategic visualization of critical data points and actionable insights to mitigate bankruptcy risks. Looking forward to connecting with fellow professionals and discussing the future of financial risk management. Your feedback and thoughts are highly appreciated! #DataScience #MachineLearning #FinancialAnalysis #BankruptcyPrevention #DataVisualization #AI #DataAnalytics #RiskManagement #BusinessIntelligence #FinancialStability #PredictiveAnalytics