Commit 07e04f56 by Willi Rath

### Move 24 hockey sticks to the front

parent 825445f8
 ... ... @@ -130,18 +130,17 @@ repeat what was done. --- class: center, middle class: middle, center ## Example — A Simple Time Series? --- class: center, middle count: false class: left, bottom .center[] _**Figure 01.**_ Annual-mean HadISST anomalies. .smaller[[This notebook][fig_01_notebook_on_nbviewer] shows the full analysis.] ??? ... ... @@ -150,6 +149,57 @@ surface temperature). --- class: left, bottom count: false .center[] .smaller[[This notebook][fig_02_notebook_on_nbviewer] details a subtlety with the _order_ of averaging.] ??? Difference arise from subtleties in treating missing data making sum, mean, etc. lose strict linearity. --- class: left, bottom count: false .right[.smaller[*... note that we’re still weighting all months equally.*]] .center[] .smaller[[This notebook][fig_03_notebook_on_nbviewer] adds arithmetic averages and shows all 12 variants.] ??? With correctly weighted months, we have 24 curves. (Most of which are arguably wrong not in no way less likely to be actually used.) There’s a very nice blog post on [_**informal descriptions**_ vs. _**executable implementations**_.][Hinsen2017] --- class: middle, center count: false .center[] .right[.smaller[© [Fabien Perissinotto](https://commons.wikimedia.org/wiki/User:Fabienp)]] ??? Don’t make the mistake of refuting this rather artificial example: - Simple time series are often used to define indices which then feed composite analyses etc. - combining only a few indices with two or three reasonable implementations each is quickly growing to `2^n` or `3^n` possible outcomes. --- class: top, left ### The Sloppy Way ... ... @@ -369,73 +419,6 @@ Suppose, this was a multi-author paper. Then, it would be easy class: middle, left ## Interlude Let’s compare different ways to calculate the SST anomalies: .center[] .right[.smaller[*... need your notes*]] --- count: false ## Interlude - Two Lines [This notebook][fig_02_notebook_on_nbviewer] details a subtlety with the _order_ of averaging: .center[] ??? Difference arise from subtleties in treating missing data making sum, mean, etc. lose strict linearity. --- count: false ## Interlude - Twelve Lines [This notebook][fig_03_notebook_on_nbviewer] adds arithmetic averages and shows all 12 variants: .center[] .right[.smaller[*... note that we’re still weighting all months equally.*]] ??? With correctly weighted months, we have 24 curves. (Most of which are arguably wrong not in no way less likely to be actually used.) There’s a very nice blog post on [_**informal descriptions**_ vs. _**executable implementations**_.][Hinsen2017] --- class: middle, center count: false .center[] .right[.smaller[© [Fabien Perissinotto](https://commons.wikimedia.org/wiki/User:Fabienp)]] ??? Don’t make the mistake of refuting this rather artificial example: - Simple time series are often used to define indices which then feed composite analyses etc. - combining only a few indices with two or three reasonable implementations each is quickly growing to `2^n` or `3^n` possible outcomes. --- class: middle, left ## Building Repeatable Work Flows 1. Provide a data set containing _**all the numbers**_ necessary to re-plot and ... ...
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!