O'Reilly on Data

DECEMBER 12, 2019

If a model is going to be used on all kinds of people, it’s best to ensure the training data has a representative distribution of all kinds of people as well. Interpretable ML models and explainable ML. The debugging techniques we propose should work on almost any kind of ML-based predictive model.

Machine Learning 191

Machine Learning Modeling Testing Risk Management 191

Domino Data Lab

APRIL 21, 2021

Rules-based fraud detection (top) vs. classification decision tree-based detection (bottom): The risk scoring in the former model is calculated using policy-based, manually crafted rules and their corresponding weights. from sklearn import metrics. from sklearn import metrics. It can be implemented as either unsupervised (e.g.

Statistics 69

Statistics Machine Learning Modeling Metrics 69

Domino Data Lab

AUGUST 22, 2019

Although it’s not perfect, [Note: These are statistical approximations, of course!] GloVe and word2vec differ in their underlying methodology: word2vec uses predictive models, while GloVe is count based. Note: A test set of 19,500 such analogies was developed by Tomas Mikolov and his colleagues in their 2013 word2vec paper.

Deep Learning 111

Deep Learning Modeling Metrics Testing 111

Jet Global

MAY 1, 2023

Companies like Tableau (which raised over $250 million when it had its IPO in 2013) demonstrated an unmet need in the market. As a result, end users can better view shared metrics (backed by accurate data), which ultimately drives performance. They can also create custom calculations and metrics, and build new data visualizations.

Analytics 52

Analytics Cost-Benefit Visualization Dashboards 52