Remove understanding-causal-inference
article thumbnail

In AI we trust? Why we Need to Talk About Ethics and Governance (part 2 of 2)

Cloudera

In part 1 of this blog post, we discussed the need to be mindful of data bias and the resulting consequences when certain parameters are skewed. Systems should be designed with bias, causality and uncertainty in mind. Even if protected features are removed, they can often be inferred from the presence of proxy features.

article thumbnail

Causal attribution in an era of big time-series data

The Unofficial Google Data Science Blog

by KAY BRODERSEN For the first time in the history of statistics, recent innovations in big data might allow us to estimate fine-grained causal effects, automatically and at scale. How can we establish that there is a causal link between our idea and the outcome metric we care about? Causal inference at scale may be one such key.

Insiders

Sign Up for our Newsletter

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

article thumbnail

Understanding Simpson’s Paradox to Avoid Faulty Conclusions

Sisense

The paradox can be resolved by better understanding the data — exploring how it was generated and identifying the lurking variable. To better understand when the data should be grouped, you should be familiar with causal inference. This drug seems to be bad for women, bad for men, but good for people!”.

Testing 104
article thumbnail

Attributing a deep network’s prediction to its input features

The Unofficial Google Data Science Blog

By MUKUND SUNDARARAJAN, ANKUR TALY, QIQI YAN Editor's note: Causal inference is central to answering questions in science, engineering and business and hence the topic has received particular attention on this blog. In this post, we explore causal inference in this setting via the problem of attribution in deep networks.

IT 70
article thumbnail

Changing assignment weights with time-based confounders

The Unofficial Google Data Science Blog

However, if one changes assignment weights when there are time-based confounders, then ignoring this complexity can lead to biased inference in an OCE. Just as in ramp-up, making inferences while ignoring the complexity of time-based confounders that are present can lead to biased estimates. Y_i=Y_i(t)$ if $T_i=t$.

article thumbnail

Causality in machine learning

The Unofficial Google Data Science Blog

We can estimate the causal effect of prominence as the difference in log odds of uptake between stories which were recommended and those which were eligible (i.e. The former approach will tend to produce more realistic outcomes, but can be harder to understand, and may not give us adequate data to assess unlikely decisions.

article thumbnail

New Applied ML Research: Meta-Learning & Structural Time Series

Cloudera

In the past year, we’ve released research reports and prototypes exploring Deep Learning for Anomaly Detection , Causality for Machine Learning and NLP for Automated Question Answering. textbooks, and understanding potential applications and limitations by prototyping. . Evolving Research At Cloudera Fast Forward.