forked from bcaffo/courses
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathindex.html
More file actions
282 lines (244 loc) · 7.99 KB
/
Copy pathindex.html
File metadata and controls
282 lines (244 loc) · 7.99 KB
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
<!DOCTYPE html>
<html>
<head>
<title>Prediction study design</title>
<meta charset="utf-8">
<meta name="description" content="Prediction study design">
<meta name="author" content="Jeffrey Leek">
<meta name="generator" content="slidify" />
<meta name="apple-mobile-web-app-capable" content="yes">
<meta http-equiv="X-UA-Compatible" content="chrome=1">
<link rel="stylesheet" href="../../librariesNew/frameworks/io2012/css/default.css" media="all" >
<link rel="stylesheet" href="../../librariesNew/frameworks/io2012/css/phone.css"
media="only screen and (max-device-width: 480px)" >
<link rel="stylesheet" href="../../librariesNew/frameworks/io2012/css/slidify.css" >
<link rel="stylesheet" href="../../librariesNew/highlighters/highlight.js/css/tomorrow.css" />
<base target="_blank"> <!-- This amazingness opens all links in a new tab. -->
<!-- Grab CDN jQuery, fall back to local if offline -->
<script src="http://ajax.aspnetcdn.com/ajax/jQuery/jquery-1.7.min.js"></script>
<script>window.jQuery || document.write('<script src="../../librariesNew/widgets/quiz/js/jquery.js"><\/script>')</script>
<script data-main="../../librariesNew/frameworks/io2012/js/slides"
src="../../librariesNew/frameworks/io2012/js/require-1.0.8.min.js">
</script>
</head>
<body style="opacity: 0">
<slides class="layout-widescreen">
<!-- LOGO SLIDE -->
<slide class="title-slide segue nobackground">
<aside class="gdbar">
<img src="../../assets/img/bloomberg_shield.png">
</aside>
<hgroup class="auto-fadein">
<h1>Prediction study design</h1>
<h2></h2>
<p>Jeffrey Leek<br/>Johns Hopkins Bloomberg School of Public Health</p>
</hgroup>
<article></article>
</slide>
<!-- SLIDES -->
<slide class="" id="slide-1" style="background:;">
<hgroup>
<h2>Prediction study design</h2>
</hgroup>
<article data-timings="">
<ol>
<li>Define your error rate</li>
<li>Split data into:
<ul>
<li>Training, Testing, Validation (optional)</li>
</ul></li>
<li>On the training set pick features
<ul>
<li>Use cross-validation</li>
</ul></li>
<li>On the training set pick prediction function
<ul>
<li>Use cross-validation</li>
</ul></li>
<li>If no validation
<ul>
<li>Apply 1x to test set</li>
</ul></li>
<li>If validation
<ul>
<li>Apply to test set and refine</li>
<li>Apply 1x to validation </li>
</ul></li>
</ol>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-2" style="background:;">
<hgroup>
<h2>Know the benchmarks</h2>
</hgroup>
<article data-timings="">
<p><img class=center src=../../assets/img/08_PredictionAndMachineLearning/benchmark.png height=400></p>
<p><a href="http://www.heritagehealthprize.com/c/hhp/leaderboard">http://www.heritagehealthprize.com/c/hhp/leaderboard</a></p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-3" style="background:;">
<hgroup>
<h2>Study design</h2>
</hgroup>
<article data-timings="">
<p><img class=center src=../../assets/img/08_PredictionAndMachineLearning/studyDesign.png height=400></p>
<p><a href="http://www2.research.att.com/%7Evolinsky/papers/ASAStatComp.pdf">http://www2.research.att.com/~volinsky/papers/ASAStatComp.pdf</a></p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-4" style="background:;">
<hgroup>
<h2>Used by the professionals</h2>
</hgroup>
<article data-timings="">
<p><img class=center src=../../assets/img/08_PredictionAndMachineLearning/kagglefront.png height=400></p>
<p><a href="http://www.kaggle.com/">http://www.kaggle.com/</a></p>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-5" style="background:;">
<hgroup>
<h2>Avoid small sample sizes</h2>
</hgroup>
<article data-timings="">
<ul>
<li>Suppose you are predicting a binary outcome
<ul>
<li>Diseased/healthy</li>
<li>Click on ad/not click on ad </li>
</ul></li>
<li>One classifier is flipping a coin</li>
<li>Probability of perfect classification is approximately:
<ul>
<li>\(\left(\frac{1}{2}\right)^{sample \; size}\)</li>
<li>\(n = 1\) flipping coin 50% chance of 100% accuracy</li>
<li>\(n = 2\) flipping coin 25% chance of 100% accuracy</li>
<li>\(n = 10\) flipping coin 0.10% chance of 100% accuracy</li>
</ul></li>
</ul>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-6" style="background:;">
<hgroup>
<h2>Rules of thumb for prediction study design</h2>
</hgroup>
<article data-timings="">
<ul>
<li>If you have a large sample size
<ul>
<li>60% training</li>
<li>20% test</li>
<li>20% validation</li>
</ul></li>
<li>If you have a medium sample size
<ul>
<li>60% training</li>
<li>40% testing</li>
</ul></li>
<li>If you have a small sample size
<ul>
<li>Do cross validation</li>
<li>Report caveat of small sample size</li>
</ul></li>
</ul>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="" id="slide-7" style="background:;">
<hgroup>
<h2>Some principles to remember</h2>
</hgroup>
<article data-timings="">
<ul>
<li>Set the test/validation set aside and <em>don't look at it</em></li>
<li>In general <em>randomly</em> sample training and test</li>
<li>Your data sets must reflect structure of the problem
<ul>
<li>If predictions evolve with time split train/test in time chunks (called<a href="http://en.wikipedia.org/wiki/Backtesting">backtesting</a> in finance)</li>
</ul></li>
<li>All subsets should reflect as much diversity as possible
<ul>
<li>Random assignment does this</li>
<li>You can also try to balance by features - but this is tricky</li>
</ul></li>
</ul>
</article>
<!-- Presenter Notes -->
</slide>
<slide class="backdrop"></slide>
</slides>
<div class="pagination pagination-small" id='io2012-ptoc' style="display:none;">
<ul>
<li>
<a href="#" target="_self" rel='tooltip'
data-slide=1 title='Prediction study design'>
1
</a>
</li>
<li>
<a href="#" target="_self" rel='tooltip'
data-slide=2 title='Know the benchmarks'>
2
</a>
</li>
<li>
<a href="#" target="_self" rel='tooltip'
data-slide=3 title='Study design'>
3
</a>
</li>
<li>
<a href="#" target="_self" rel='tooltip'
data-slide=4 title='Used by the professionals'>
4
</a>
</li>
<li>
<a href="#" target="_self" rel='tooltip'
data-slide=5 title='Avoid small sample sizes'>
5
</a>
</li>
<li>
<a href="#" target="_self" rel='tooltip'
data-slide=6 title='Rules of thumb for prediction study design'>
6
</a>
</li>
<li>
<a href="#" target="_self" rel='tooltip'
data-slide=7 title='Some principles to remember'>
7
</a>
</li>
</ul>
</div> <!--[if IE]>
<script
src="http://ajax.googleapis.com/ajax/libs/chrome-frame/1/CFInstall.min.js">
</script>
<script>CFInstall.check({mode: 'overlay'});</script>
<![endif]-->
</body>
<!-- Load Javascripts for Widgets -->
<!-- MathJax: Fall back to local if CDN offline but local image fonts are not supported (saves >100MB) -->
<script type="text/x-mathjax-config">
MathJax.Hub.Config({
tex2jax: {
inlineMath: [['$','$'], ['\\(','\\)']],
processEscapes: true
}
});
</script>
<script type="text/javascript" src="http://cdn.mathjax.org/mathjax/2.0-latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script>
<!-- <script src="https://c328740.ssl.cf1.rackcdn.com/mathjax/2.0-latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML">
</script> -->
<script>window.MathJax || document.write('<script type="text/x-mathjax-config">MathJax.Hub.Config({"HTML-CSS":{imageFont:null}});<\/script><script src="../../librariesNew/widgets/mathjax/MathJax.js?config=TeX-AMS-MML_HTMLorMML"><\/script>')
</script>
<!-- LOAD HIGHLIGHTER JS FILES -->
<script src="../../librariesNew/highlighters/highlight.js/highlight.pack.js"></script>
<script>hljs.initHighlightingOnLoad();</script>
<!-- DONE LOADING HIGHLIGHTER JS FILES -->
</html>