add modeling-02

covariates.qmd

@@ -79,7 +79,7 @@ We need to create a random sample of background in both time and space.
 
 ### Sampling time
 
-Sampling time requires us to consider that the occurrences are not even distributed through time.  We can see that using a histogram of observation dates by month.
+Sampling time requires us to consider that the occurrences are not evenly distributed through time.  We can see that using a histogram of observation dates by month.
 
 ```{r}
 H = hist(obs$date, breaks = 'month', format = "%Y", 
@@ -120,6 +120,10 @@ days_sample = sample(days, size = nback, replace = TRUE, prob = day_probs)
 
 So, now we have a sampling of of dates that have a temporal distribution similar to that of the observations.
 
+:::{.callout-warning}
+It is possible that we maybe [overfitting](https://en.wikipedia.org/wiki/Overfitting) by weighting the samples in time. Other time-sampling strategies are available to us, so we are not stuck with the approach we are using and can easily revisit this selection.
+:::
+
 ### Sampling space
 
 The [sf](https://CRAN.R-project.org/package=sf) package provides a function, `st_sample()`, for sampling points within a polygon.  But what polygon?  We have choices as we could use (a) a bounding box around the observations, (b) a convex hull around the observations or (c) a buffered envelope around the observations. Each has it's advantages and disadvantages.  We show how to make one of each.

docs/covariates.html

@@ -320,7 +320,7 @@ <h2 data-number="3" class="anchored" data-anchor-id="sampling-background-data"><
 <p>We need to create a random sample of background in both time and space.</p>
 <section id="sampling-time" class="level3" data-number="3.1">
 <h3 data-number="3.1" class="anchored" data-anchor-id="sampling-time"><span class="header-section-number">3.1</span> Sampling time</h3>
-<p>Sampling time requires us to consider that the occurrences are not even distributed through time. We can see that using a histogram of observation dates by month.</p>
+<p>Sampling time requires us to consider that the occurrences are not evenly distributed through time. We can see that using a histogram of observation dates by month.</p>
 <div class="cell" data-hash="covariates_cache/html/unnamed-chunk-3_6c57817ed29ec1b0d03b0ce25e6fe886">
 <div class="sourceCode cell-code" id="cb4"><pre class="sourceCode r code-with-copy"><code class="sourceCode r"><span id="cb4-1"><a href="#cb4-1" aria-hidden="true" tabindex="-1"></a>H <span class="ot">=</span> <span class="fu">hist</span>(obs<span class="sc">$</span>date, <span class="at">breaks =</span> <span class="st">'month'</span>, <span class="at">format =</span> <span class="st">"%Y"</span>, </span>
 <span id="cb4-2"><a href="#cb4-2" aria-hidden="true" tabindex="-1"></a>     <span class="at">freq =</span> <span class="cn">TRUE</span>, <span class="at">main =</span> <span class="st">"Observations"</span>,</span>
@@ -363,6 +363,19 @@ <h3 data-number="3.1" class="anchored" data-anchor-id="sampling-time"><span clas
 <span id="cb8-3"><a href="#cb8-3" aria-hidden="true" tabindex="-1"></a>days_sample <span class="ot">=</span> <span class="fu">sample</span>(days, <span class="at">size =</span> nback, <span class="at">replace =</span> <span class="cn">TRUE</span>, <span class="at">prob =</span> day_probs)</span></code><button title="Copy to Clipboard" class="code-copy-button"><i class="bi"></i></button></pre></div>
 </div>
 <p>So, now we have a sampling of of dates that have a temporal distribution similar to that of the observations.</p>
+<div class="callout callout-style-default callout-warning callout-titled">
+<div class="callout-header d-flex align-content-center">
+<div class="callout-icon-container">
+<i class="callout-icon"></i>
+</div>
+<div class="callout-title-container flex-fill">
+Warning
+</div>
+</div>
+<div class="callout-body-container callout-body">
+<p>It is possible that we maybe <a href="https://en.wikipedia.org/wiki/Overfitting">overfitting</a> by weighting the samples in time. Other time-sampling strategies are available to us, so we are not stuck with the approach we are using and can easily revisit this selection.</p>
+</div>
+</div>
 </section>
 <section id="sampling-space" class="level3" data-number="3.2">
 <h3 data-number="3.2" class="anchored" data-anchor-id="sampling-space"><span class="header-section-number">3.2</span> Sampling space</h3>

docs/index.html

@@ -332,9 +332,22 @@ <h3 data-number="3.4" class="anchored" data-anchor-id="data-storage"><span class
 │   ├── bkg-covariates.gpkg
 │   └── buffered-polygon.gpkg
 ├── model
-│   └── v1
-│       └── v1.0
-│           └── model_v1.0.rds
+│   ├── v1
+│   │   └── v1.0
+│   │       └── model_v1.0.rds
+│   └── v2
+│       ├── v2.Apr.rds
+│       ├── v2.Aug.rds
+│       ├── v2.Dec.rds
+│       ├── v2.Feb.rds
+│       ├── v2.Jan.rds
+│       ├── v2.Jul.rds
+│       ├── v2.Jun.rds
+│       ├── v2.Mar.rds
+│       ├── v2.May.rds
+│       ├── v2.Nov.rds
+│       ├── v2.Oct.rds
+│       └── v2.Sep.rds
 ├── nbs
 │   ├── 2000
 │   │   ├── 0101
@@ -1472,7 +1485,6 @@ <h3 data-number="3.4" class="anchored" data-anchor-id="data-storage"><span class
 │   │       ├── uvcomp_SCIMonthlyGlobal.v_wind.2023-07-01.tif
 │   │       └── uvcomp_SCIMonthlyGlobal.windspeed.2023-07-01.tif
 │   └── database.csv.gz
-├── nbs2
 ├── obis
 │   └── Mola_mola.gpkg
 ├── obs

docs/modeling-01.html

@@ -310,7 +310,7 @@ <h2 data-number="5" class="anchored" data-anchor-id="save-the-model"><span class
 <p>In this tutorial we are going to build three types of models: basic, monthly and split (between testing and training). We should organize the storage of the models in a way that makes sense. With each model we may generate one or more predictions - for example, for our basic model we might tyr to hind-cast a individual years. That’s a one-to-many to relationship between model and predictions. We suggest that you start considering each model a version and store them accordingly. Let’s use a simple numbering scheme…</p>
 <ul>
 <li><code>v1.0</code> for the basic model</li>
-<li><code>v2.01, v2.01, ..., v2.12</code> for the monthly models</li>
+<li><code>v2.jan, v2.feb, ..., v2.dec</code> for the monthly models</li>
 <li><code>v3.0, ...</code> for for the split model(s)</li>
 </ul>
 <p>The <a href="https://github.com/BigelowLab/maxnetic">maxnetic</a> provides some convenience functions for working with maxnet models including file storage functions.</p>