Bird Collisions

Bootstrapping

Published

May 2, 2019

Notable topics: Bootstrapping

Recorded on: 2019-05-02

Timestamps by: Alex Cookson

View code

Screencast

Timestamps

Analyzing when NAs appear in a dimension

gather

Looking at multiple categorical variable at the same time by gathering them into one column and eventually graphing each as a different facet

Re-order facet graphs according to which ones have the fewest categories in them to ones that have the most

Geometric mean for estimating counts when there are a lot of low values (1-3 bird collisions, in this case)

Filling in "blank" observations where there were no observations made

Using log+1 to convert a dimension with values of 0 into a log scale

Adding confidence bounds for data using a geometric mean (where he first gets the idea of bootstrapping)

Actual coding of bootstrap starts

Adding confidence bounds using bootstrap data

Investigating potential confounding variables

Discussing approaches to dealing with confounding variables

complete

Using complete function to get explicit NA values