article thumbnail

The Lean Analytics Cycle: Metrics > Hypothesis > Experiment > Act

Occam's Razor

To win in business you need to follow this process: Metrics > Hypothesis > Experiment > Act. We are far too enamored with data collection and reporting the standard metrics we love because others love them because someone else said they were nice so many years ago. That metric is tied to a KPI.

Metrics 156
article thumbnail

Towards optimal experimentation in online systems

The Unofficial Google Data Science Blog

the weight given to Likes in our video recommendation algorithm) while $Y$ is a vector of outcome measures such as different metrics of user experience (e.g., Crucially, it takes into account the uncertainty inherent in our experiments. Here, $X$ is a vector of tuning parameters that control the system's operating characteristics (e.g.

Insiders

Sign Up for our Newsletter

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

article thumbnail

Measuring Validity and Reliability of Human Ratings

The Unofficial Google Data Science Blog

Once we’ve answered that, we will then define and use metrics to understand the quality of human-labeled data, along with a measurement framework that we call Cross-replication Reliability or xRR. We will follow the example of Janson and Olsson , and start from this generalized definition of the metric, which they call iota.

article thumbnail

Why model calibration matters and how to achieve it

The Unofficial Google Data Science Blog

Calibration and other considerations Calibration is a desirable property, but it is not the only important metric. bar{pi} (1 - bar{pi})$: This is the irreducible loss due to uncertainty. Other important losses we consider are accuracy (the proportion of correct classifications) and discrimination based metrics like AUC.

Modeling 122
article thumbnail

Estimating causal effects using geo experiments

The Unofficial Google Data Science Blog

This means it is possible to specify exactly in which geos an ad campaign will be served – and to observe the ad spend and the response metric at the geo level. In other words, iROAS is the slope of a curve of the response metric plotted against the underlying advertising spend. Cambridge, 2007. by turning campaigns off).