The Unofficial Google Data Science Blog

APRIL 23, 2024

If the relationship of $X$ to $Y$ can be approximated as quadratic (or any polynomial), the objective and constraints as linear in $Y$, then there is a way to express the optimization as a quadratically constrained quadratic program (QCQP). Crucially, it takes into account the uncertainty inherent in our experiments.

Experimentation 52

Experimentation Optimization Uncertainty Metrics 52

The Unofficial Google Data Science Blog

MARCH 31, 2016

In the context of prediction problems, another benefit is that the models produce an estimate of the uncertainty in their predictions: the predictive posterior distribution. Often our data can be stored or visualized as a table like the one shown below. Cambridge University Press, (2006). [2] bandit problems). ICML, (2005). [3]

Modeling 40

Modeling Statistics Advertising Testing 40

Domino Data Lab

JULY 22, 2019

He also really informed a lot of the early thinking about data visualization. It involved a lot of work with applied math, some depth in statistics and visualization, and also a lot of communication skills. The problems down in the mature bucket, those are optimizations, they aren’t showstoppers. How could that make sense?

Data Science 109

Data Science Machine Learning Data Governance Modeling 109

The Unofficial Google Data Science Blog

JULY 11, 2017

If both variances are positive then the optimal estimator of $y_{t+1}$ winds up being "exponential smoothing," where past data are forgotten at an exponential rate determined by the ratio of the two variances. There are also plotting functions that you can use to visualize the regression coefficients. Compare to Figure 2. Gramacy, R.

Forecasting 64

Forecasting Statistics Modeling Software 64