Kurtosis is a statistical measure that quantifies the "tailedness" of a probability distribution, indicating how much of the data is concentrated in the tails relative to a normal distribution. If you're a data scientist at a company like Facebook or Google, you might encounter kurtosis when analyzing user behavior data or optimizing machine learning models.

How to use it in a sentence

• As a data scientist at a startup, I calculated the kurtosis of our user engagement metrics and found that the distribution had heavy tails, suggesting that a few power users were driving most of our traffic while the majority of users were relatively inactive. Maybe we should pivot to a new product that appeals to a broader audience and not just the tech elite.

• I was optimizing a deep learning model for image recognition and noticed that the kurtosis of the activation values in the hidden layers was extremely high, indicating that the network was relying on a small number of neurons to make predictions. I guess I'll have to add some regularization to prevent overfitting, even though it means I'll be stuck debugging this model all weekend instead of going to that hackathon.

If you actually want to learn more...

• K-means clustering is not a free lunch – Variance Explained by David Robinson, a data scientist at Heap, explores the assumptions behind the k-means clustering algorithm and how violating these assumptions can lead to misleading results. http://varianceexplained.org/r/kmeans-free-lunch/

• How to interpret a p-value histogram – Variance Explained, also by David Robinson, discusses how to interpret different shapes of p-value histograms and what they reveal about the performance of statistical tests. http://varianceexplained.org/statistics/interpreting-pvalue-histogram/

• Modeling gene expression with broom: a case study in tidy analysis – Variance Explained showcases how to use the broom package in R to analyze gene expression data and identify genes with interesting patterns, such as outliers or nutrient-specific responses. http://varianceexplained.org/r/tidy-genomics-broom/

Note: the Developer Dictionary is in Beta. Please direct feedback to skye@statsig.com.