-
Notifications
You must be signed in to change notification settings - Fork 0
/
ll-3-overview-presentation.html
561 lines (401 loc) · 15.9 KB
/
ll-3-overview-presentation.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
<!DOCTYPE html>
<html lang="" xml:lang="">
<head>
<title>Machine Learning Learning Lab 3</title>
<meta charset="utf-8" />
<meta name="author" content="Dr. Joshua Rosenberg" />
<meta name="date" content="2022-07-01" />
<script src="libs/header-attrs-2.14/header-attrs.js"></script>
<link href="libs/remark-css-0.0.1/default.css" rel="stylesheet" />
<link href="libs/panelset-0.2.6/panelset.css" rel="stylesheet" />
<script src="libs/panelset-0.2.6/panelset.js"></script>
<script src="libs/clipboard-2.0.6/clipboard.min.js"></script>
<link href="libs/xaringanExtra-clipboard-0.2.6/xaringanExtra-clipboard.css" rel="stylesheet" />
<script src="libs/xaringanExtra-clipboard-0.2.6/xaringanExtra-clipboard.js"></script>
<script>window.xaringanExtraClipboard(null, {"button":"<i class=\"fa fa-clipboard\"><\/i>","success":"<i class=\"fa fa-check\" style=\"color: #90BE6D\"><\/i>","error":"Press Ctrl+C to Copy"})</script>
<link href="libs/font-awesome-5.1.0/css/all.css" rel="stylesheet" />
<link href="libs/font-awesome-5.1.0/css/v4-shims.css" rel="stylesheet" />
<link href="libs/tile-view-0.2.6/tile-view.css" rel="stylesheet" />
<script src="libs/tile-view-0.2.6/tile-view.js"></script>
<link rel="stylesheet" href="css/laser.css" type="text/css" />
<link rel="stylesheet" href="css/laser-fonts.css" type="text/css" />
</head>
<body>
<textarea id="source">
class: clear, title-slide, inverse, center, top, middle
# Machine Learning Learning Lab 3
----
### **Dr. Joshua Rosenberg**
### July 01, 2022
---
# Background
- Once we've processed our variables in new ways and have made our model better, we want to select the best model
- Best models can be understood in two ways:
- tuning
- selection
- Focus: selecting a model, setting tuning parameters, and picking the best performing model
- We'll focus on random forest models as the context for this, though the points apply to any complex algorithm
---
# Agenda
.pull-left[
## Part 1: Core Concepts
- tuning
- model selection
]
.pull-right[
## Part 2: R Code-Along
- NGSS and transactional and substantive for one year (again)
- Different ways of estimating tuning parameters
]
---
class: clear, inverse, center, middle
# Core Concepts
---
# How do I select a model?
One general principle is to **start with the simplest useful model** and to _build toward
more complex models as helpfuL_.
This principle applies in multiple ways:
- To choose an algorithm, start with simpler models that you can efficiently use and understand
- To carry out feature engineering, understand your predictors well by starting with a subset
- To tune an algorithm, start with a relatively simple set of tuning parameters
This isn't just for beginners or those of us in education; [most spam filters use Support Vector Machines (and used Naive Bayes until recently)](https://vas3k.com/blog/machine_learning/) due to their combination of effectiveness and efficiency "in production."
---
# The bias-variance tradeoff
- An important way to achieve good performance with test data is to balance between the inherent _bias_ in your algorithm and the _variance_ in the predictions of your algorithm; this is referred to as the **bias-variance** trade-off of _all_ models
.footnote[
[1] not always/often used, for reasons we'll discuss later
]
---
# Illustrating the bias-variance tradeoff
<img src="ll-3-overview-presentation_files/figure-html/unnamed-chunk-2-1.png" style="display: block; margin: auto;" />
---
# Strong bias
<img src="ll-3-overview-presentation_files/figure-html/unnamed-chunk-3-1.png" style="display: block; margin: auto;" />
---
# A much less-biased algorithm
<img src="ll-3-overview-presentation_files/figure-html/unnamed-chunk-4-1.png" style="display: block; margin: auto;" />
---
# Slightly different data (right pane)
.pull-left[
<img src="img/bias-variance-data-1.png" width=400 height=400>
]
.pull-right[
<img src="img/bias-variance-data-2.png" width=400 height=400>
]
---
# Still strong bias, but low variance
.pull-left[
<img src="img/bias-variance-data-3.png" width=400 height=400>
]
.pull-right[
<img src="img/bias-variance-data-4.png" width=400 height=400>
]
---
# Low bias, but very high variance
.pull-left[
<img src="img/bias-variance-data-5.png" width=400 height=400>
]
.pull-right[
<img src="img/bias-variance-data-6.png" width=400 height=400>
]
---
# The bias-variance tradeoff
.pull-left[
#### Bias
- *Definition*: Difference between our known codes/outcomes and our predicted codes/outcomes; difference between `\(y\)` and `\(\hat{y}\)`
- How (in)correct our models' (algorithms') predictions are
- Models with high bias can fail to capture important relationships--they can be *under-fit* to our data
- In short, how well our model reflects the patterns in the data
]
.pull-right[
#### Variance
- *Definition*: Using a different sample of data, the difference in `\(\hat{y}\)` values
- How sensitive our predictions are to the specific sample on which we trained the model
- Models with high variance can fail to predict different data well--they can be *over-fit* to our data
- In short, how stable the predictions of our model are
<h4><center>Regardless of model, we often wish to balance between bias and variance</center></h4>
---
# Tuning
- Many parts of models - their _parameters_ are estimated from the data
- Other parts cannot be estimated from the data and must be used:
- They are often set as defaults
- But you can often improve on these defaults
---
# Aside: Random forests and classification trees
- *Random forests* are extensions of classification trees
- _Classification trees_ are a type of algorithm that - at their core - use conditional logic ("if-then" statements) in a _nested_ manner
- For instance, here's a _very, very_ simple tree (from [APM](https://link.springer.com/book/10.1007/978-1-4614-6849-3)):
- Measures are used to determine the splits in such a way that classifies observations into small, homogeneous groups (using measures such as the Gini index and entropy measure)
---
# A more complex tree
As you can imagine, with many variables, these trees can become very complex
---
# Random forests
- Random forest is an extension of decision tree modeling, whereby a collection of decision trees are simultaneously estimated ("grown") and are evaluated based on out-of-sample predictive accuracy
- Random forest estimates all the decision trees at once so each tree is independent of every other tree.
- The random forest algorithm provides a regression approach that is distinct from other modeling approaches.
- The final random forest model aggregates the findings across all the separate trees in the forest
---
<!-- -->
[Koehrsen (2017)](https://williamkoehrsen.medium.com/random-forest-simple-explanation-377895a60d2d)
---
# Tuning parameters for random forests
- There are several important tuning parameters for these models:
- the number of predictor columns that are randomly sampled for each split in the tree (`mtry` in the code we'll run later)
- the number of data points required to execute a split (`min_n`)
- the minimum number of observations in each node (`size`)
- These tuning parameters, broadly, balance predictive performance with the training data with how well the model will perform on new data
---
# Overview of regression modeling
1. Split data (into train, test, and tuning sets sets)
1. Specify recipe, model, and workflow
1. Estimate models and determine tuning parameters' values
1. Evaluate metrics and predictions
---
class: clear, inverse, center, middle
# Code Examples
---
.panelset[
.panel[.panel-name[1]
**Split data**
```r
library(tidymodels) # doesn't load forcats, stringr, readr from tidyverse
library(readr)
library(here)
library(vip)
d <- read_csv(here("spring-workshop", "data-to-model.csv"))
d <- select(d, -time_spent) # this is another continuous outcome
train_test_split <- initial_split(d, prop = .70)
data_train <- training(train_test_split)
kfcv <- vfold_cv(data_train)
```
]
.panel[.panel-name[2]
**Engineer features**
```r
# pre-processing/feature engineering
# d <- select(d, student_id:final_grade, subject:percomp) # selecting the contextual/demographic variables
# and the survey variables
d <- d %>% select(-student_id)
sci_rec <- recipe(final_grade ~ ., data = d) %>%
add_role(course_id, new_role = "ID variable") %>% # this can be any string
step_novel(all_nominal_predictors()) %>%
step_normalize(all_numeric_predictors()) %>%
step_dummy(all_nominal_predictors()) %>%
step_nzv(all_predictors()) %>%
step_impute_knn(all_predictors(), all_outcomes())
```
]
.panel[.panel-name[3]
**Specify recipe, model, and workflow**
```r
# specify model
rf_mod_many <-
rand_forest(mtry = tune(),
min_n = tune()) %>%
set_engine("ranger", importance = "impurity") %>%
set_mode("regression")
# specify workflow
rf_wf_many <-
workflow() %>%
add_model(rf_mod_many) %>%
add_recipe(sci_rec)
```
]
.panel[.panel-name[4]
**Fit model**
```r
# specify tuning grid
finalize(mtry(), data_train)
finalize(min_n(), data_train)
tree_grid <- grid_max_entropy(mtry(range(1, 15)),
min_n(range(2, 40)),
size = 10)
# fit model with tune_grid
tree_res <- rf_wf_many %>%
tune_grid(
resamples = kfcv,
grid = tree_grid,
metrics = metric_set(rmse, mae, rsq)
)
```
]
.panel[.panel-name[5]
**Fit model (part 2)**
```r
# examine best set of tuning parameters; repeat?
show_best(tree_res, n = 10)
# select best set of tuning parameters
best_tree <- tree_res %>%
select_best()
# finalize workflow with best set of tuning parameters
final_wf <- rf_wf_many %>%
finalize_workflow(best_tree)
# fit split data (separately)
final_fit <- final_wf %>%
last_fit(train_test_split, metrics = metric_set(rmse, mae, rsq))
```
]
.panel[.panel-name[6]
**Evaluate accuracy**
```r
# variable importance plot
final_fit %>%
pluck(".workflow", 1) %>%
pull_workflow_fit() %>%
vip(num_features = 10)
# fit stats
final_fit %>%
collect_metrics()
# test set predictions
final_fit %>%
collect_predictions()
```
]
]
---
# We'll next dive deeper
- **Guided walkthrough**: work through this code to tune a model
- **Independent practice**: try tuning a neural network
- **Readings**: tuning an advanced model - deep learning
---
class: clear, center
## .font130[.center[**Thank you!**]]
<br/>
.center[<img style="border-radius: 80%;" src="img/jr-cycling.jpeg" height="200px"/><br/>**Dr. Joshua Rosenberg**<br/><mailto:[email protected]>]
</textarea>
<style data-target="print-only">@media screen {.remark-slide-container{display:block;}.remark-slide-scaler{box-shadow:none;}}</style>
<script src="https://remarkjs.com/downloads/remark-latest.min.js"></script>
<script>var slideshow = remark.create({
"highlightStyle": "default",
"highlightLines": true,
"highlightLanguage": "r",
"countIncrementalSlides": false,
"ratio": "16:9",
"slideNumberFormat": "<div class=\"progress-bar-container\">\n <div class=\"progress-bar\" style=\"width: calc(%current% / %total% * 100%);\">\n </div>\n</div>"
});
if (window.HTMLWidgets) slideshow.on('afterShowSlide', function (slide) {
window.dispatchEvent(new Event('resize'));
});
(function(d) {
var s = d.createElement("style"), r = d.querySelector(".remark-slide-scaler");
if (!r) return;
s.type = "text/css"; s.innerHTML = "@page {size: " + r.style.width + " " + r.style.height +"; }";
d.head.appendChild(s);
})(document);
(function(d) {
var el = d.getElementsByClassName("remark-slides-area");
if (!el) return;
var slide, slides = slideshow.getSlides(), els = el[0].children;
for (var i = 1; i < slides.length; i++) {
slide = slides[i];
if (slide.properties.continued === "true" || slide.properties.count === "false") {
els[i - 1].className += ' has-continuation';
}
}
var s = d.createElement("style");
s.type = "text/css"; s.innerHTML = "@media print { .has-continuation { display: none; } }";
d.head.appendChild(s);
})(document);
// delete the temporary CSS (for displaying all slides initially) when the user
// starts to view slides
(function() {
var deleted = false;
slideshow.on('beforeShowSlide', function(slide) {
if (deleted) return;
var sheets = document.styleSheets, node;
for (var i = 0; i < sheets.length; i++) {
node = sheets[i].ownerNode;
if (node.dataset["target"] !== "print-only") continue;
node.parentNode.removeChild(node);
}
deleted = true;
});
})();
// add `data-at-shortcutkeys` attribute to <body> to resolve conflicts with JAWS
// screen reader (see PR #262)
(function(d) {
let res = {};
d.querySelectorAll('.remark-help-content table tr').forEach(tr => {
const t = tr.querySelector('td:nth-child(2)').innerText;
tr.querySelectorAll('td:first-child .key').forEach(key => {
const k = key.innerText;
if (/^[a-z]$/.test(k)) res[k] = t; // must be a single letter (key)
});
});
d.body.setAttribute('data-at-shortcutkeys', JSON.stringify(res));
})(document);
(function() {
"use strict"
// Replace <script> tags in slides area to make them executable
var scripts = document.querySelectorAll(
'.remark-slides-area .remark-slide-container script'
);
if (!scripts.length) return;
for (var i = 0; i < scripts.length; i++) {
var s = document.createElement('script');
var code = document.createTextNode(scripts[i].textContent);
s.appendChild(code);
var scriptAttrs = scripts[i].attributes;
for (var j = 0; j < scriptAttrs.length; j++) {
s.setAttribute(scriptAttrs[j].name, scriptAttrs[j].value);
}
scripts[i].parentElement.replaceChild(s, scripts[i]);
}
})();
(function() {
var links = document.getElementsByTagName('a');
for (var i = 0; i < links.length; i++) {
if (/^(https?:)?\/\//.test(links[i].getAttribute('href'))) {
links[i].target = '_blank';
}
}
})();
// adds .remark-code-has-line-highlighted class to <pre> parent elements
// of code chunks containing highlighted lines with class .remark-code-line-highlighted
(function(d) {
const hlines = d.querySelectorAll('.remark-code-line-highlighted');
const preParents = [];
const findPreParent = function(line, p = 0) {
if (p > 1) return null; // traverse up no further than grandparent
const el = line.parentElement;
return el.tagName === "PRE" ? el : findPreParent(el, ++p);
};
for (let line of hlines) {
let pre = findPreParent(line);
if (pre && !preParents.includes(pre)) preParents.push(pre);
}
preParents.forEach(p => p.classList.add("remark-code-has-line-highlighted"));
})(document);</script>
<script>
slideshow._releaseMath = function(el) {
var i, text, code, codes = el.getElementsByTagName('code');
for (i = 0; i < codes.length;) {
code = codes[i];
if (code.parentNode.tagName !== 'PRE' && code.childElementCount === 0) {
text = code.textContent;
if (/^\\\((.|\s)+\\\)$/.test(text) || /^\\\[(.|\s)+\\\]$/.test(text) ||
/^\$\$(.|\s)+\$\$$/.test(text) ||
/^\\begin\{([^}]+)\}(.|\s)+\\end\{[^}]+\}$/.test(text)) {
code.outerHTML = code.innerHTML; // remove <code></code>
continue;
}
}
i++;
}
};
slideshow._releaseMath(document);
</script>
<!-- dynamically load mathjax for compatibility with self-contained -->
<script>
(function () {
var script = document.createElement('script');
script.type = 'text/javascript';
script.src = 'https://mathjax.rstudio.com/latest/MathJax.js?config=TeX-MML-AM_CHTML';
if (location.protocol !== 'file:' && /^https?:/.test(script.src))
script.src = script.src.replace(/^https?:/, '');
document.getElementsByTagName('head')[0].appendChild(script);
})();
</script>
</body>
</html>