forked from perlatex/R_for_Data_Science
-
Notifications
You must be signed in to change notification settings - Fork 0
/
eda_penguins.Rmd
615 lines (439 loc) · 14.7 KB
/
eda_penguins.Rmd
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
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
# 探索性数据分析-企鹅的故事 {#eda-penguins}
今天讲一个关于企鹅的数据故事。这个故事来源于科考人员记录的大量企鹅体征[数据](https://raw.githubusercontent.com/rfordatascience/tidytuesday/master/data/2020/2020-07-28/penguins.csv),图片来源[这里](https://github.com/allisonhorst/palmerpenguins).
```{r eda-penguins-1, out.width = '100%', echo = FALSE}
knitr::include_graphics("images/penguins.png")
```
## 数据
### 导入数据
可通过宏包`palmerpenguins::penguins`获取数据,也可以读取本地`penguins.csv`文件,
我们采取后面一种方法:
```{r eda-penguins-2, eval=FALSE, include=FALSE}
library(tidyverse)
d <- palmerpenguins::penguins
d %>%
tidyr::drop_na() %>%
head()
```
```{r eda-penguins-3, message = FALSE, warning = FALSE}
library(tidyverse)
penguins <- read_csv("./demo_data/penguins.csv") %>%
janitor::clean_names()
penguins %>%
head()
```
### 变量含义
|variable |class |description |
|:-----------------|:-------|:-----------|
|species |integer | 企鹅种类 (Adelie, Gentoo, Chinstrap) |
|island |integer | 所在岛屿 (Biscoe, Dream, Torgersen) |
|bill_length_mm |double | 嘴峰长度 (单位毫米) |
|bill_depth_mm |double | 嘴峰深度 (单位毫米)|
|flipper_length_mm |integer | 鰭肢长度 (单位毫米) |
|body_mass_g |integer | 体重 (单位克) |
|sex |integer | 性别 |
|year |integer | 记录年份 |
```{r eda-penguins-4, out.width = '86%', echo = FALSE}
knitr::include_graphics("images/culmen_depth.png")
```
### 数据清洗
检查缺失值(NA)这个很重要!
```{r eda-penguins-5}
penguins %>% summarise(
across(everything(), ~ sum(is.na(.)))
)
```
有缺失值的地方找出来看看
```{r eda-penguins-6}
penguins %>% filter_all(
any_vars(is.na(.))
)
```
发现共有11行至少有一处有缺失值,于是我们就删除这些行
```{r eda-penguins-7}
penguins <- penguins %>% drop_na()
penguins
```
## 探索性分析
大家可以提出自己想探索的内容:
- 每种类型企鹅有多少只?
- 每种类型企鹅各种属性的均值和分布?
- 嘴峰长度和深度的关联?
- 体重与翅膀长度的关联?
- 嘴峰长度与嘴峰深度的比例?
- 不同种类的宝宝,体重具有显著性差异?
- 这体征中哪个因素对性别影响最大?
- ...
### 每种类型企鹅有多少只
```{r eda-penguins-8}
penguins %>%
count(species, sort = T)
```
### 每个岛屿有多少企鹅?
```{r eda-penguins-9}
penguins %>%
count(island, sort = T)
```
### 每种类型企鹅各种体征属性的均值和分布
```{r eda-penguins-10}
penguins %>%
group_by(species) %>%
summarize(across(where(is.numeric), mean, na.rm = TRUE))
```
### 每种类型企鹅的嘴峰长度的分布
```{r eda-penguins-11}
penguins %>%
ggplot(aes(x = bill_length_mm)) +
geom_density() +
facet_wrap(vars(species), scales = "free")
```
### 每种类型企鹅的嘴峰长度的分布(分性别)
```{r eda-penguins-12}
penguins %>%
ggplot(aes(x = bill_length_mm)) +
geom_density(aes(fill = sex)) +
facet_wrap(vars(species), scales = "free")
```
男宝宝的嘴巴要长些,哈哈。
来张更好看点的
```{r eda-penguins-13}
penguins %>%
ggplot(aes(x = bill_length_mm, fill = sex)) +
geom_histogram(
position = "identity",
alpha = 0.7,
bins = 25
) +
scale_fill_manual(values = c("#66b3ff", "#8c8c8c")) +
ylab("number of penguins") +
xlab("length (mm)") +
theme_minimal() +
theme(
legend.position = "bottom",
legend.text = element_text(size = 11),
legend.title = element_blank(),
panel.grid.minor = element_blank(),
axis.title = element_text(color = "white", size = 10),
plot.title = element_text(size = 20),
plot.subtitle = element_text(size = 12, hjust = 1)
) +
facet_wrap(vars(species), scales = "free")
```
同理,可以画出其他属性的分布。当然,我更喜欢用山峦图来呈现不同分组的分布,因为竖直方向可以更方便比较
```{r eda-penguins-14}
library(ggridges)
penguins %>%
ggplot(aes(x = bill_length_mm, y = species, fill = species)) +
ggridges::geom_density_ridges()
```
同样,我们也用颜色区分下性别,这样不同种类、不同性别企鹅的嘴峰长度分布一目了然
```{r eda-penguins-15}
penguins %>%
ggplot(aes(x = bill_length_mm, y = species, fill = sex)) +
geom_density_ridges(alpha = 0.5)
```
同样的代码,类似地画个其他体征的分布,
```{r eda-penguins-16}
penguins %>%
ggplot(aes(x = bill_depth_mm, fill = species)) +
ggridges::geom_density_ridges(aes(y = species))
```
```{r eda-penguins-17}
penguins %>%
ggplot(aes(x = bill_depth_mm, fill = sex)) +
ggridges::geom_density_ridges(aes(y = species))
```
```{r eda-penguins-18}
penguins %>%
ggplot(aes(x = body_mass_g, y = species, fill = sex)) +
ggridges::geom_density_ridges(alpha = 0.5)
```
但这样一个特征一个特征的画,好麻烦。你知道程序员都是偷懒的,于是我们还有更骚的操作
```{r eda-penguins-19}
penguins %>%
dplyr::select(species, bill_length_mm:body_mass_g) %>%
pivot_longer(-species, names_to = "measurement", values_to = "value") %>%
ggplot(aes(x = value)) +
geom_density(aes(color = species, fill = species), size = 1.2, alpha = 0.2) +
facet_wrap(vars(measurement), ncol = 2, scales = "free")
```
```{r eda-penguins-20}
penguins %>%
dplyr::select(species, bill_length_mm:body_mass_g) %>%
pivot_longer(-species, names_to = "measurement", values_to = "value") %>%
ggplot(aes(x = species, y = value)) +
geom_boxplot(aes(color = species, fill = species), size = 1.2, alpha = 0.2) +
facet_wrap(vars(measurement), ncol = 2, scales = "free")
```
```{r eda-penguins-21}
penguins %>%
dplyr::select(species, bill_length_mm:body_mass_g) %>%
pivot_longer(-species, names_to = "measurement", values_to = "value") %>%
ggplot(aes(x = value, y = species, fill = species)) +
ggridges::geom_density_ridges() +
facet_wrap(vars(measurement), scales = "free")
```
```{r eda-penguins-22}
penguins %>%
dplyr::select(species,sex, bill_length_mm:body_mass_g) %>%
pivot_longer(
-c(species, sex),
names_to = "measurement",
values_to = "value"
) %>%
ggplot(aes(x = value, y = species, fill = sex)) +
ggridges::geom_density_ridges() +
facet_wrap(vars(measurement), scales = "free")
```
我若有所思的看着这张图,似乎看到了一些特征(pattern)了。
### 嘴峰长度和深度的关联
嘴巴越长,嘴巴也会越厚?
```{r eda-penguins-23}
penguins %>%
ggplot(aes(
x = bill_length_mm, y = bill_depth_mm,
shape = species, color = species
)) +
geom_point()
```
我们把不同的种类,用不同的颜色区分看看
```{r eda-penguins-24}
penguins %>%
ggplot(aes(
x = bill_length_mm, y = bill_depth_mm,
shape = species, color = species
)) +
geom_point(aes(size = body_mass_g))
```
感觉这是一个辛普森佯谬, 我们画图看看
```{r eda-penguins-25}
penguins %>%
ggplot(aes(x = bill_length_mm, y = bill_depth_mm)) +
geom_point(aes(color = species, shape = species)) +
geom_smooth(method = lm) +
geom_smooth(method = lm, aes(color = species))
```
### 体重与翅膀长度的关联
翅膀越长,体重越大?
```{r eda-penguins-26}
penguins %>%
group_by(species, island, sex) %>%
ggplot(aes(
x = body_mass_g, y = reorder(species, -body_mass_g),
color = species
)) +
geom_jitter(position = position_jitter(seed = 2020, width = 0.2), alpha = 0.4, size = 2) +
stat_summary(fun = mean, geom = "point", size = 5, alpha = 1)
```
```{r eda-penguins-27}
library(ggtext)
penguins %>%
ggplot(aes(flipper_length_mm, body_mass_g, group = species)) +
geom_point(aes(colour = species, shape = species), alpha = 0.7) +
scale_color_manual(values = c("darkorange", "purple", "cyan4")) +
labs(
title = "Penguin Size, Palmer Station LTER",
subtitle = "Flipper length and body mass for <span style = 'color:darkorange;'>Adelie</span>, <span style = 'color:purple;'>Chinstrap</span> and <span style = 'color:cyan4;'>Gentoo</span> Penguins",
x = "flipper length (mm)",
y = "body mass (g)"
) +
theme_minimal() +
theme(
legend.position = "none",
# text = element_text(family = "Futura"),
# (I only have 'Light' )
plot.title = element_text(size = 16),
plot.subtitle = element_markdown(), # element_markdown from `ggtext` to parse the css in the subtitle
plot.title.position = "plot",
plot.caption = element_text(size = 8, colour = "grey50"),
plot.caption.position = "plot"
)
```
### 不同种类的宝宝,体重具有显著性差异?
先分组计算体重的均值和标准差
```{r eda-penguins-28}
penguins %>%
group_by(species) %>%
summarise(
count = n(),
mean_body_mass = mean(body_mass_g),
sd_body_mass = sd(body_mass_g)
)
```
```{r eda-penguins-29}
penguins %>%
ggplot(aes(x = species, y = body_mass_g)) +
geom_boxplot() +
geom_jitter()
```
用统计方法验证下我们的猜测吧。记住,我们是有科学精神的的人!
#### 参数检验
- one-way ANOVA(要求等方差)
```{r eda-penguins-30}
stats::aov(formula = body_mass_g ~ species, data = penguins) %>%
summary()
```
p-value 很小,说明不同种类企鹅之间体重是有显著差异的,但aov只给出了species在整体上引起了体重差异(只要有任意两组之间有显著差异,aov给出的p-value都很小),如果想知道不同种类两两之间是否有显著差异,这就需要用到TukeyHSD().
- one-way ANOVA(不要求等方差),相关介绍看[here](http://www.sthda.com/english/wiki/one-way-anova-test-in-r)
```{r eda-penguins-31}
oneway.test(body_mass_g ~ species, data = penguins)
```
```{r eda-penguins-32}
stats::aov(formula = body_mass_g ~ species, data = penguins) %>%
TukeyHSD(which = "species") %>%
broom::tidy()
```
表格第一行instrap-Adelie 的 p-value = 0.916,没通过显著性检验;而Gentoo-Adelie 和 Gentoo-Chinstrap 他们的p-value都接近0,通过显著性检验,这和图中的结果是一致的。
作为统计出生的R语言,有很多宏包可以帮助我们验证我们的结论,我这里推荐**可视化学统计**的宏包[ggstatsplot](https://indrajeetpatil.github.io/ggstatsplot/)宏包将统计分析的结果写在图片里,统计结果和图形融合在一起,让统计结果更容易懂了。(使用这个宏包辅助我们学习统计)
```{r eda-penguins-33, eval=FALSE}
library(ggstatsplot)
penguins %>%
ggstatsplot::ggbetweenstats(
x = species, # > 2 groups
y = body_mass_g,
type = "parametric",
pairwise.comparisons = TRUE,
pairwise.display = "all",
messages = FALSE,
var.equal = FALSE
)
```
#### 非参数检验
相关介绍看[here](http://www.sthda.com/english/wiki/kruskal-wallis-test-in-r)
```{r eda-penguins-34}
kruskal.test(body_mass_g ~ species, data = penguins)
```
```{r eda-penguins-35, eval=FALSE}
penguins %>%
ggstatsplot::ggbetweenstats(
x = species,
y = body_mass_g,
type = "nonparametric",
mean.ci = TRUE,
pairwise.comparisons = TRUE, # <<
pairwise.display = "all", # ns = only non-significant
p.adjust.method = "fdr", # <<
messages = FALSE
)
```
哇,原来统计可以这样学!
### 嘴峰长度与嘴峰深度的比例
```{r eda-penguins-36}
penguins %>%
mutate(ratio = bill_length_mm / bill_depth_mm) %>%
group_by(species) %>%
summarise(mean = mean(ratio))
```
```{r eda-penguins-37}
penguins %>%
mutate(ratio = bill_length_mm / bill_depth_mm) %>%
ggplot(aes(x = ratio, fill = species)) +
ggridges::geom_density_ridges(aes(y = species))
```
男宝宝和女宝宝颜色区分下,代码只需要修改一个地方,留给大家自己实践下吧。
### 建立模型
建模需要标准化数据,并对分类变量(比如sex)编码为 1 和 0; (这是第二个好习惯)
```{r eda-penguins-38}
scale_fun <- function(x) {
(x - mean(x)) / sd(x)
}
d <- penguins %>%
select(sex, species, bill_length_mm:body_mass_g) %>%
mutate(
across(where(is.numeric), scale_fun)
) %>%
mutate(male = if_else(sex == "male", 1, 0))
d
```
按照species分组后,对flipper_length_mm标准化?这样数据会聚拢到一起了喔, 还是不要了
```{r eda-penguins-39, eval=FALSE}
penguins %>%
select(sex, species, bill_length_mm:body_mass_g) %>%
group_by(species) %>%
mutate(
across(where(is.numeric), scale_fun)
) %>%
ungroup()
```
#### model_01
我们将性别sex视为响应变量,其他变量为预测变量。这里性别变量是二元的(0 或者 1),所以我们用logistic回归
```{r eda-penguins-40}
logit_mod1 <- glm(
male ~ 1 + species + bill_length_mm + bill_depth_mm +
flipper_length_mm + body_mass_g,
data = d,
family = binomial(link = "logit")
)
summary(logit_mod1)
```
计算每个变量的平均边际效应
```{r eda-penguins-41}
library(margins)
logit_mod1_m <- logit_mod1 %>%
margins() %>%
summary() %>%
as_tibble()
logit_mod1_m
```
```{r eda-penguins-42}
logit_mod1_m %>%
ggplot(aes(
x = reorder(factor, AME),
y = AME, ymin = lower, ymax = upper
)) +
geom_hline(yintercept = 0, color = "gray80") +
geom_pointrange() +
coord_flip() +
labs(x = NULL, y = "Average Marginal Effect")
```
```{r eda-penguins-43, eval=FALSE}
library(ggeffects)
ggpredict(logit_mod1, terms = "bill_length_mm")
```
#### model_02
```{r eda-penguins-44, eval=FALSE}
library(brms)
brms_mod2 <- brm(
male ~ 1 + bill_length_mm + bill_depth_mm + flipper_length_mm + body_mass_g + (1 | species),
data = d,
family = binomial(link = "logit")
)
```
```{r eda-penguins-45, eval=FALSE}
summary(brms_mod2)
```
```{r eda-penguins-46, eval=FALSE}
library(ggeffects)
ggpredict(brms_mod2, "bill_depth_mm [all]") %>%
plot()
```
#### model_03
```{r eda-penguins-47, eval=FALSE}
penguins %>%
ggplot(aes(x = flipper_length_mm, y = bill_length_mm, color = species)) +
geom_point()
```
```{r eda-penguins-48, eval=FALSE}
brms_mod3 <- brm(bill_length_mm ~ flipper_length_mm + (1|species),
data = penguins
)
```
```{r eda-penguins-49, eval=FALSE}
penguins %>%
group_by(species) %>%
modelr::data_grid(flipper_length_mm) %>%
tidybayes::add_fitted_draws(brms_mod3, n = 100) %>%
ggplot() +
geom_point(
data = penguins,
aes(flipper_length_mm, bill_length_mm, color = species, shape = species)
) +
geom_line(aes(flipper_length_mm, .value, group = interaction(.draw, species), color = species), alpha = 0.1)
```
```{r eda-penguins-50, echo = F}
# remove the objects
# rm(list=ls())
rm(d, logit_mod1, logit_mod1_m, penguins, scale_fun)
```
```{r eda-penguins-51, echo = F, message = F, warning = F, results = "hide"}
pacman::p_unload(pacman::p_loaded(), character.only = TRUE)
```