-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathExploratory Data Analysis.Rmd
1417 lines (1156 loc) · 56.2 KB
/
Exploratory Data Analysis.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
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
---
title: "WildFires Analysis - Exploratory Data Analysis"
output:
html_document:
df_print: paged
html_notebook: default
pdf_document: default
---
```{r warning=FALSE}
# Loading packages
library(pryr) # Memory usage information
library(lubridate) # date manipulation
library(dplyr) # data manipulation
library(tidyr) # data manipulation
library(ggplot2) # graphics
library(directlabels) # End lines legend for ggplot line plots
library(plotly) # Interactive graphics
library(stringr) # string manipulation
library(stringi) # remove diacritics
library(ggfortify)
library(rjson) # read json
library(nasapower) # nasapower api data
library(ggsci) # Scientific color palletes
library(zoo) # Replace NAs in time series analysis
library(rgdal) # Work with geo-spatial data
library(sf) # geo-spatial vector data
library(TSA) # Time Series Analys
library(qwraps2)# for pretty summary statistics
```
There are 4 csv files containing data about the fires of all ocurrences around all brazilian territory. Each file has data between 1st Sept to 30th Sept of each year from 2015 to 2019.
```{r}
temp <- list.files(path = "./database/modis_15-19", pattern = "*.csv")
myfiles = lapply(paste0("./database/modis_15-19/", temp), read.csv)
data <- bind_rows(myfiles)
rm(temp)
rm(myfiles)
```
there are 885117 observations in our dataset which sums up to 70 MB of storaged data.
```{r}
str(data)
```
```{r}
summary(data)
```
In order to print a well formatted table of summary statistics, lets write a function to do it
```{r}
options(qwraps2_markup = "latex")
my_summary <-
list("diasemchuva" =
list("min" = ~ min(.data$diasemchuva, na.rm = T),
"mediana" = ~ qwraps2::median_iqr(.data$diasemchuva, na_rm = T),
"média (desv. padr.)" = ~ qwraps2::mean_sd(.data$diasemchuva, na_rm = T),
"máx" = ~ max(.data$diasemchuva, na.rm = T)),
"precipitacao" =
list("min" = ~ min(.data$precipitacao, na.rm = T),
"mediana" = ~ qwraps2::median_iqr(.data$precipitacao, na_rm = T),
"média (desv. padr.)" = ~ qwraps2::mean_sd(.data$precipitacao, na_rm = T),
"máx" = ~ max(.data$precipitacao, na.rm = T)),
"riscofogo" =
list("min" = ~ min(.data$riscofogo, na.rm = T),
"mediana" = ~ qwraps2::median_iqr(.data$riscofogo, na_rm = T),
"média (desv. padr.)" = ~ qwraps2::mean_sd(.data$riscofogo, na_rm = T),
"máx" = ~ max(.data$riscofogo, na.rm = T)),
"frp" =
list("min" = ~ min(.data$frp, na.rm = T),
"mediana" = ~ qwraps2::median_iqr(.data$frp, na_rm = T),
"média (desv. padr.)" = ~ qwraps2::mean_sd(.data$frp, na_rm = T),
"máx" = ~ max(.data$frp, na.rm = T)),
"Bioma" =
list("Amazônia" = ~ qwraps2::n_perc0(.data$bioma == 'Amazonia'),
"Cerrado" = ~ qwraps2::n_perc0(.data$bioma == 'Cerrado'),
"Mata Atlântica" = ~ qwraps2::n_perc0(.data$bioma == 'Mata Atlantica'),
"Pantanal" = ~ qwraps2::n_perc0(.data$bioma == 'Pantanal'),
"Caatinga" = ~ qwraps2::n_perc0(.data$bioma == 'Caatinga'),
"Pampa" = ~ qwraps2::n_perc0(.data$bioma == 'Pampa')
)
)
### Overall - output is a latex text
pretty_summary <- summary_table(data, my_summary)
```
Aparentemente, A coluna de área industrial e FRP tem poucas entradas não nulas
```{r}
total_records <- nrow(data)
noWeatherInfo <- is.na(data$diasemchuva)
noFrpInfo <- is.na(data$frp)
paste("Percentage of NAs entries on `FRP` =",
round(100 * sum(noFrpInfo) / total_records, 2), "%")
paste("Percentage of NAs entries on `diasemchuva` =",
round(100 * (sum(noWeatherInfo)) / total_records, 2), "%")
```
By the last results, it is possible that the number of days with no rain have a huge impact over the fires triggers.
we can also investigate another columns that have either redundant data or not useful data.
```{r}
levels(data$pais)
levels(data$satelite)
data$pais <- NULL
data$satelite <- NULL
object_size(data)
```
By looking at the structure of our dataframe, it becomes clear that the `datahora` is a datetime object not a Factor.
```{r}
data$datahora <- ymd_hms(str_replace_all(data$datahora,"/","-"))
levels(data$bioma)
```
We have 6 distincts biomas on our dataset. Perhaps, by knowing how is the distribution of fires among different biomas, we can refine our research.
```{r}
counts <- table(data$bioma)
counts
proportions <- 100 * counts / sum(counts)
proportions
```
As we can see, Amazonia represents around half of our dataset. In addition, the number of fire occurrences is very much like the proportion of land that each bioma has in brazilian territory.
-------------------------- Visualisations -----------------------------
```{r, fig.width = 4, fig.height= 3}
# For text in plots see
# http://r-statistics.co/Complete-Ggplot2-Tutorial-Part2-Customizing-Theme-With-R-Code.html
# For legend in plots see
# https://www.datanovia.com/en/blog/ggplot-legend-title-position-and-labels/
path_ch4 <- "/home/alexandre/Documentos/tcc/images/plots_cap4/"
proportions <- data %>% group_by(bioma) %>% summarise(n = round(100*n()/nrow(.), 1))
quantities <- as.character(sort(desc(proportions$n)))
ggplot(proportions, aes(x = reorder(bioma, -n), y = n, fill = bioma)) +
geom_bar(stat = "identity") +
coord_flip() +
scale_fill_manual(values = c("#ff5a00", "#999999", "#ffce00", "#999999", "#999999", "#999999")) +
labs(title="Proporção de observações por Bioma", y="Proporção %", x="Bioma", caption="Fonte: Elaborado pelo Autor") +
theme(plot.title=element_text(size=10,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=10,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title.x=element_text(size=8), # X axis title
axis.title.y=element_text(size=8), # Y axis title
axis.text.x=element_text(size=6,
vjust=.5), # X axis text
axis.text.y=element_text(size=6),
legend.title = element_blank(),
legend.position = "none",
plot.caption = element_text(size = 6)) # Y axis text
# ggsave(paste0(path_ch4, "bioma-proportions.png"), dpi = "retina", width = 10, height = 8, units = "cm")
```
----------------------- DETECTING SEASONALITY
```{r}
data %>%
mutate(date = date(datahora)) %>%
group_by(date) %>%
ggplot(aes(x = date, group = 1)) +
geom_freqpoly(stat = "count") +
scale_x_date(date_breaks = "6 months", date_labels = "%b/%y") +
labs(y = "Quantidade de incêndios", x = "Mês/Ano", title = 'Quantidade de incêndios por dia', caption="Fonte: Elaborado pelo Autor") +
theme(plot.title=element_text(size=10,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=10,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title.x=element_text(size=8), # X axis title
axis.title.y=element_text(size=8), # Y axis title
axis.text.x=element_text(size=6,
vjust=.5), # X axis text
axis.text.y=element_text(size=6),
legend.title = element_blank(),
legend.position = "none",
plot.caption = element_text(size = 8)) # Y axis text
# ggsave(paste0(path_ch4, "count-byDay.png"), dpi = "retina", width = 10, height = 8, units = "cm")
```
As we can see above, our data presents strong patterns (seasonality). By doing a Fourier Transform, it is possible get to know what are the main frequencies of this time series.
```{r}
sinal_INPE <- data %>%
mutate(date = date(datahora)) %>%
group_by(date) %>%
summarise(n = n())
p <- periodogram(sinal_INPE$n, plot = F)
dd = data.frame(freq=p$freq, spec=p$spec)
order = dd[order(-dd$spec),]
top3 = head(order, 3)
# frequências com 3 maiores potências
top3
```
```{r}
time <- 1/top3$freq
time
```
```{r}
freqs <- dd %>% arrange(desc(dd$spec)) %>% filter(freq <= 0.1)
freqs %>% ggplot(mapping = aes(x = freq, y = spec)) +
geom_bar(stat = "identity") +
labs(y = "Potência", x = "Frequência (Hz)", title = 'Densidade Espectral de Potência', caption="Fonte: Elaborado pelo Autor") +
theme(plot.title=element_text(size=10,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=10,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title.x=element_text(size=8), # X axis title
axis.title.y=element_text(size=8), # Y axis title
axis.text.x=element_text(size=6,
vjust=.5), # X axis text
axis.text.y=element_text(size=6),
legend.title = element_blank(),
legend.position = "none",
plot.caption = element_text(size = 8)) # Y axis text
# ggsave(paste0(path_ch4, "psd.png"), dpi = "retina", width = 10, height = 8, units = "cm")
```
Therefore, the seasonality of our dataset is given by annualy, semesterly and quaterly periods. For reference, see https://anomaly.io/detect-seasonality-using-fourier-transform-r/index.html and http://www.di.fc.ul.pt/~jpn/r/fourier/fourier.html
Now, it is also important to get the underlying trend of the number of fire ocurrences through the months of the year. Let's start exploring this idea:
```{r}
# Capitalizes names
capitalize <- function(x) {
s <- strsplit(x, " ")[[1]]
paste(toupper(substring(s, 1,1)), substring(s, 2),
sep="", collapse=" ")
}
levels <- c("Jan", "Fev", "Mar", "Abr", "Mai", "Jun", "Jul", "Ago", "Set", "Out", "Nov", "Dez")
temp <- data %>%
transmute(mes = factor(sapply(substr(months(data$datahora), 1, 3), capitalize), levels = levels),
ano = factor(year(datahora))) %>%
group_by(ano, mes)
cold_dfs <- list()
for (i in 1:5) {
cold_dfs[[i]] <- temp %>%
filter(!mes %in% c("Jan", "Fev", "Mar", "Abr", "Mai", "Dez"), ano == (i + 2014))
cold_dfs[[i]]$color <- paste0("cold", i)
}
hot_dfs <- list()
for (i in 1:5) {
hot_dfs[[i]] <- temp %>%
filter(!mes %in% c("Jun", "Jul", "Ago", "Set", "Out", "Nov"), ano == (i + 2014))
hot_dfs[[i]]$color <- paste0("hot", i)
}
colors_df <- bind_rows(cold_dfs, hot_dfs) %>%
group_by(ano, mes, color) %>%
summarise(n = n()) %>%
mutate(ano_abb = as.factor(as.numeric(as.character(ano)) %% 2000))
dodge <- position_dodge(width = 0.9)
colors_df %>%
ggplot(aes(x = mes, y = n)) +
geom_bar(stat = "identity", position = dodge, aes(fill = color)) +
# geom_text(aes(label = ano_abb), position = dodge,
# size = 1.5, inherit.aes = T) +
scale_colour_manual("Ano", values = c(cold1 = "#b7a8c1", cold2 = "#c7b4cc",
cold3 ="#d7c0d7", cold4 = "#e7cce2", cold5 = "#f7d8ed",
hot1 = "#ff4949", hot2 = "#ff6549",
hot3 ="#ff8049", hot4 = "#ff9c49", hot5 = "#ffb749")) +
labs(y = "Quantidade de incêndios", x = "Mês", title = 'Quantidade mensal de incêndios', subtitle = "em cada ano", caption="Fonte: Elaborado pelo Autor") +
theme(plot.title=element_text(size=10,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=8,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title.x=element_text(size=8), # X axis title
axis.title.y=element_text(size=8), # Y axis title
axis.text.x=element_text(size=6,
vjust=.5), # X axis text
axis.text.y=element_text(size=6),
legend.title = element_blank(),
legend.position = "none",
plot.caption = element_text(size = 8)) + # Y axis text
ggsave(paste0(path_ch4, "count-byMonth.png"), dpi = "retina", width = 10, height = 8, units = "cm")
rm(temp)
# rm(colors_df)
```
```{r}
data %>%
mutate(month = floor_date(date(datahora), "month")) %>%
group_by(month, bioma) %>%
summarise(n_fires = n()) %>%
ggplot(aes(x = month, y = n_fires/1000)) +
geom_bar(aes(fill = bioma), stat = 'identity') +
scale_x_date(date_labels = "%b/%y", date_breaks = "6 months") +
labs(y = "Focos de incêndios (milhares)", x = "Mês/Ano", title = 'Participação de cada Bioma em incêndios/mês', caption="Fonte: Elaborado pelo Autor") +
guides(fill = guide_legend(title = "Bioma", override.aes = list(size = 0.5))) +
theme(plot.title=element_text(size=9,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=9,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title.x=element_text(size=7), # X axis title
axis.title.y=element_text(size=7), # Y axis title
axis.text.x=element_text(size=5,
vjust=.5), # X axis text
axis.text.y=element_text(size=5),
legend.position=c(1,1),
legend.justification=c(0, 1),
legend.key.width= unit(0.4, "lines"),
legend.key.height = unit(0.4, "lines"),
plot.margin = unit(c(1, 5, 0.5, 0.5), "lines"),
legend.title = element_text(size = 4),
legend.text = element_text(size = 3),
plot.caption = element_text(size = 6)) + # Y axis text
ggsave(paste0(path_ch4, "trends-byMonthbyBiomabyYear.png"), dpi = "retina", width = 10, height = 8, units = "cm")
```
In order to plot the number of fires by state, we must do some data wrangling to put the data in the desired format.
```{r}
is.odd <- function(x) x %% 2 != 0
is.even <- function(x) x %% 2 == 0
geo_data <- fromJSON(file = "./geo_data/geo_data_states.json")
maps_df <- list()
for (i in 1:length(geo_data$features)) {
coordinates <- unlist(geo_data$features[[i]]$geometry$coordinates[[1]])
long <- coordinates[is.odd(seq_along(coordinates))]
lat <- coordinates[is.even(seq_along(coordinates))]
if(length(long) != length(lat)){
stop("Wrong subset of coordinate vector", call. = FALSE)
}
geo_length <- length(long)
estado <- toupper(geo_data$features[[i]]$properties$name)
codigo <- as.integer(geo_data$features[[i]]$properties$codigo_ibg)
regiao <- as.integer(geo_data$features[[i]]$properties$regiao_id)
maps_df[[i]] <- as_tibble(
list(
estado = rep(stri_trans_general(estado,"Latin-ASCII"), geo_length),
codigo = rep(codigo, geo_length),
regiao = rep(regiao, geo_length),
longitude = long,
latitude = lat
)
)
}
map_df <- bind_rows(maps_df)
```
```{r}
# the `estado` column of the map dataframe is a character, so in order
# to avoid coertion from factor to character, let's convert it explicitly
data$estado <- as.character(data$estado)
```
```{r}
data %>%
select(estado) %>%
group_by(estado) %>%
summarise(n = n()) %>%
right_join(map_df, by = 'estado') %>%
ggplot(aes(x = longitude, y = latitude, group = codigo, fill = n/1000)) +
geom_polygon() +
geom_path(color = 'white', size = 0.1) +
scale_fill_continuous(low = "orange",
high = "darkred",
name = 'Nº de incêndios (milhares)'
) +
labs(title = 'Incêndios por Estado', subtitle = "Janeiro/2015 a Setembro/2019", caption="Fonte: Elaborado pelo Autor" ) +
theme(plot.title=element_text(size=9,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=7,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title = element_blank(),
axis.text = element_blank(),
axis.ticks = element_blank(),
legend.position=c(1,1),
legend.justification=c(0, 1),
legend.key.width= unit(0.4, "lines"),
legend.key.height = unit(0.4, "lines"),
plot.margin = unit(c(1, 5, 0.5, 0.5), "lines"),
legend.title = element_text(size = 4),
legend.text = element_text(size = 3),
plot.caption = element_text(size = 6)) +
ggsave(paste0(path_ch4, "map-byState.png"), dpi = "retina", width = 10, height = 8, units = "cm")
```
As we can see, the state with the most fire ocurrences is Pará in which most of its lands belongs to the amazon ecossystem, this may explain why half the dataset ocurrences comes from Amazon fires.
As we can see below, this trend happens all years.
```{r}
byYear_df <- data %>%
select(estado, datahora) %>%
mutate(ano = year(datahora)) %>%
group_by(estado, ano) %>%
summarise(n = n()) %>%
right_join(map_df, by = 'estado')
byYear_df %>%
ggplot(aes(x = longitude, y = latitude, group = codigo, fill = n/1000)) +
geom_polygon() +
facet_wrap(~ano) +
geom_path(color = 'white', size = 0.1) +
scale_fill_continuous(low = "orange",
high = "darkred",
name = 'Números de incêndios\n(milhares)'
) +
labs(title = 'Incêndios por Estado a cada ano', caption="Fonte: Elaborado pelo Autor" ) +
theme(plot.title=element_text(size=9,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=7,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title = element_blank(),
axis.text = element_blank(),
axis.ticks = element_blank(),
legend.position=c(0.7, 0.5),
legend.justification=c(0, 1),
legend.key.width= unit(0.6, "lines"),
legend.key.height = unit(0.6, "lines"),
legend.title = element_text(size = 6),
legend.text = element_text(size = 4),
plot.caption = element_text(size = 6))
# ggsave(paste0(path_ch4, "map-byStateEachYear.pngx"), dpi = "retina", width = 10, height = 8, units ="cm")
rm(byYear_df)
```
```{r}
top10 <- data %>%
group_by(municipio, estado) %>%
summarise(n = n()) %>%
top_n(1, n) %>%
arrange(desc(n))
top10 <- top10[1:10,]
top10 %>%
ggplot(aes(x = reorder(municipio, -n), y = n, fill = estado)) +
geom_bar(stat = "identity") +
geom_text(aes(label = n),
size = 2, inherit.aes = T, hjust = 2) +
coord_flip() +
labs(title="10 municípios com maior ocorrência de focos de incêndio", subtitle = "em todo o período de Jan/15 a Set/19", y="Nº de incêndios", x= "Município", caption="Fonte: Elaborado pelo Autor") +
guides(fill = guide_legend(title = "Estado", override.aes = list(size = 0.5))) +
theme(plot.title=element_text(size=10,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=8,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title.x=element_text(size=8), # X axis title
axis.title.y=element_text(size=8), # Y axis title
axis.text.x=element_text(size=5,
vjust=.5), # X axis text
axis.text.y=element_text(size=4),
legend.justification=c(0, 1),
legend.key.width= unit(0.6, "lines"),
legend.key.height = unit(0.6, "lines"),
legend.title = element_text(size = 6),
legend.text = element_text(size = 4),
legend.position = c(0.7, 0.9),
plot.caption = element_text(size = 6)) + # Y axis text
ggsave(paste0(path_ch4, "count-byCounty.png"), dpi = "retina", width = 12, height = 8, units = "cm")
```
```{r}
# Sampling 25 percent of the whole data set to make some exploratory graphs
sample <- sample_frac(data, 0.25)
```
```{r}
data %>%
select(riscofogo, diasemchuva, precipitacao) %>%
filter(data$diasemchuva != 0, data$riscofogo != 0, data$precipitacao != 0) %>%
ggplot(aes(x=riscofogo, y=precipitacao) ) +
stat_density_2d(aes(fill = ..density..), geom = "raster", contour = F) +
scale_fill_distiller(palette=1, direction=1) +
scale_x_continuous(expand = c(0, 0)) +
scale_y_continuous(expand = c(0, 0)) +
ylim(0, 50) +
labs(title="Densidade por área", subtitle = "risco de fogo vs. precipitação (mm)", y="Precipitação (mm)", x= "Indíce de risco de fogo", caption="Fonte: Elaborado pelo Autor") +
theme(plot.title=element_text(size=10,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=8,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title.x=element_text(size=8), # X axis title
axis.title.y=element_text(size=8), # Y axis title
axis.text.x=element_text(size=5,
vjust=.5), # X axis text
axis.text.y=element_text(size=4),
legend.title = element_blank(),
legend.text = element_text(size = 4),
legend.justification=c(0, 1),
legend.key.width= unit(0.6, "lines"),
legend.key.height = unit(0.6, "lines"),
legend.position = c(0.9, 0.6),
plot.caption = element_text(size = 6)) + # Y axis text
ggsave(paste0(path_ch4, "density-PrecipRiscoFogo.png"), dpi = "retina", width = 12, height = 8, units = "cm")
```
```{r}
data %>%
ggplot(aes(x = riscofogo, y = precipitacao)) +
geom_point(alpha = 1/300, na.rm = T ) +
ylim(0, 50) +
labs(title="Gráfico de dispersão", subtitle = "risco de fogo vs. precipitação (mm)", y="Precipitação (mm)", x= "Indíce de risco de fogo", caption="Fonte: Elaborado pelo Autor") +
theme(plot.title=element_text(size=10,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=8,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title.x=element_text(size=8), # X axis title
axis.title.y=element_text(size=8), # Y axis title
axis.text.x=element_text(size=5,
vjust=.5), # X axis text
axis.text.y=element_text(size=4),
legend.title = element_blank(),
legend.text = element_text(size = 4),
legend.justification=c(0, 1),
legend.key.width= unit(0.6, "lines"),
legend.key.height = unit(0.6, "lines"),
legend.position = c(0.9, 0.6),
plot.caption = element_text(size = 6)) + # Y axis text
ggsave(paste0(path_ch4, "scatter-PrecRiscoFogo.png"), dpi = "retina", width = 12, height = 8, units = "cm")
```
In terms of raw numbers, we can visualize this difference below.
```{r}
data %>%
group_by(estado) %>%
ggplot(aes(x = factor(estado))) +
geom_bar() +
coord_flip()
```
Going down one level, let's try an analysis by county
```{r}
geo_data_counties <- fromJSON(file = "./geo_data/geo_data_counties.json")
maps_df_counties <- list()
for (i in 1:length(geo_data_counties$features)) {
coordinates <- unlist(geo_data_counties$features[[i]]$geometry$coordinates[[1]])
long <- coordinates[is.odd(seq_along(coordinates))]
lat <- coordinates[is.even(seq_along(coordinates))]
if(length(long) != length(lat)){
stop("Wrong subset of coordinate vector", call. = FALSE)
}
geo_length <- length(long)
municipio <- geo_data_counties$features[[i]]$properties$name
municipio <- toupper(stri_trans_general(municipio, "Latin-ASCII"))
codigo <- as.integer(geo_data_counties$features[[i]]$properties$id)
maps_df_counties[[i]] <- as_tibble(
list(
municipio = rep(stri_trans_general(municipio, "Latin-ASCII"), geo_length),
codigo = rep(codigo, geo_length),
longitude = long,
latitude = lat
)
)
}
map_df_county <- bind_rows(maps_df_counties)
```
```{r}
data %>%
select(municipio) %>%
group_by(municipio) %>%
summarise(n = n()) %>%
right_join(map_df_county, by = 'municipio') %>%
ggplot(aes(x = longitude, y = latitude, group = codigo, fill = n)) +
geom_polygon() +
geom_path(color = 'white', size = 0.1) +
scale_fill_continuous(low = "orange",
high = "darkred",
name = 'Number of fires'
)
```
What about that total number of fire ocurrences by state? From when does it come from?
```{r}
data %>%
mutate(ano = year(datahora)) %>%
group_by(estado, ano) %>%
ggplot(aes(x = factor(estado), fill = factor(ano))) +
geom_bar() +
coord_flip()
```
Let's repeat the last plot, but now considering the proportions
```{r}
data %>%
mutate(ano = year(datahora)) %>%
group_by(estado, ano) %>%
ggplot(aes(x = factor(estado), fill = factor(ano))) +
geom_bar(position = "fill") +
coord_flip()
```
Let's reinforce what we've seen by far and plot the data considering a monthly frequency
```{r}
levels <- c("Janeiro", "Fevereiro", "Março", "Abril", "Maio", "Junho", "Julho", "Agosto", "Setembro", "Outubro", "Novembro", "Dezembro")
data %>%
mutate(mes = factor(sapply(months(datahora), capitalize), levels = levels)) %>%
group_by(mes) %>%
ggplot(aes(x = factor(mes), y = 100*..prop.., group = 1)) +
geom_bar()
data %>%
mutate(mes = factor(sapply(months(datahora), capitalize), levels = levels),
ano = factor(year(datahora))) %>%
group_by(ano, mes) %>%
ggplot(aes(x = mes, y = 100*..prop.., group = 1)) +
geom_bar() +
facet_wrap(~ano) + coord_flip()
```
As we can see, in 2019 the trend was a bit different, maybe it is so because of the popular protests against the liberal and progressive view of brazilian government towards amazonia.
Strangely, our dataset only has records of fire occurences from 3 to 6 pm. Why is it?
```{r}
data %>%
mutate(hour = hour(datahora)) %>%
group_by(hour) %>%
ggplot(aes(x = hour, group = 1)) +
geom_bar(stat = "count")
```
O dataframe original não tem informações sobre macro regiões. Utilizando o data frame de mapas, podemos obter esta informação.
```{r}
regionsList <- list()
codMacroRegions <- list()
n_reg <- length(unique(map_df$regiao))
regions <- unique(map_df$regiao)
for(i in 1:n_reg){
temp <- map_df %>% filter(regiao == regions[i])
states <- unique(temp$estado)
regionsList[[i]] <- states
codMacroRegions[[i]] <- regions[i]
}
names(regionsList) <- c("NORTE", "NORDESTE", "SUDESTE", "CENTRO-OESTE", "SUL")
for(i in 1:n_reg){
data$macroRegiao[data$estado %in% regionsList[[i]]] <- names(regionsList)[i]
data$codMacroRegiao[data$estado %in% regionsList[[i]]] <- codMacroRegions[[i]]
}
```
```{r}
# O nome regiao do dataframe map_df nao tem o mesmo sentido que no dataframe df
# sendo assim, criou-se um dataframe auxiliar com os nomes correspondentes
temp <- data[,c("macroRegiao", "codMacroRegiao")]
names(temp) <- c("macroregiao", "regiao")
temp %>%
group_by(macroregiao, regiao) %>%
summarise(n = n()) %>%
right_join(map_df, by = 'regiao') %>%
ggplot(aes(x = longitude, y = latitude, group = regiao, fill = n/1000)) +
geom_polygon() +
geom_path(color = 'white', size = 0.1) +
scale_fill_continuous(low = "orange",
high = "darkred",
name = 'Number of fires\n(thousands)'
)
```
```{r}
na.omit(data) %>% ggplot(aes(x = frp, y = precipitacao, alpha = riscofogo)) +
geom_point() +
xlim(0,500)
na.omit(data) %>% ggplot(aes(x = log(frp), y = precipitacao, alpha = riscofogo)) +
geom_point()
```
Notice the relationship among the fire radiative power, the precipitation and the risk of fire. It seems that, with more precipitation, the frp is reduced while with less precipitation the risk of fire remains mostly
```{r}
na.omit(data) %>% ggplot(aes(x = riscofogo, y = precipitacao, alpha = frp)) +
geom_point(position = "jitter")
```
Looking at the distribution of the fire radiative power across different biomas, we can see that it is roughly the same, except for Mata Atlantica and Pampa maybe because Pampa is situated at the south brazil, where temperatures are lower which decreases the chances of fire ocurrences. On the other side, Mata Atlantica is mostly distributed along the coast, implying that the humidity that comes from the ocean may hold fire ocurrences from stroke.
```{r}
na.omit(data) %>% ggplot(aes(x = bioma, y = frp)) +
geom_boxplot()
```
```{r}
# link for reference https://jfly.uni-koeln.de/color/
# The palette with grey:
cbp1 <- c("#D55E00", "#E69F00", "#56B4E9", "#009E73",
"#F0E442", "#0072B2", "#999999", "#CC79A7")
na.omit(data) %>% ggplot(aes(x = log(frp), col = bioma, fill = bioma)) +
geom_density(alpha = 0.1) +
scale_color_manual(values = cbp1) +
scale_fill_manual(values = cbp1)
```
As we can see below, we cannot take any conclusion about the risk of fire with respect to the fire radiative power. So sad.
```{r}
na.omit(data) %>%
ggplot(aes(x = riscofogo, y = log(frp), col = bioma)) +
geom_point() + # Copy from Plot 1
geom_smooth(method = "lm", se = TRUE) +
geom_smooth(aes(group = 1), method = "lm", se = TRUE, linetype = 2)
```
One thing we did not show yet was the geographical distribution of fire points through the years. Put on your belts and enjoy the folowing two plots!
```{r}
# Segmenting Data To plot accordingly to the period available for 2019
data15 <- data %>% filter(year(datahora) == '2015')
data16 <- data %>% filter(year(datahora) == '2016')
data17 <- data %>% filter(year(datahora) == '2017')
data18 <- data %>% filter(year(datahora) == '2018')
data19 <- data %>% filter(year(datahora) == '2019')
data_partial <- data %>%
filter(month(datahora) >= 1,
month(datahora) <= 9) %>%
mutate(ano = year(datahora))
```
In addition, let's focus on the legal amazon area
```{r}
shape <- read_sf(dsn = "./geo_data/legal_amazon_simple", layer = "UF_AMLeg_LLwgs84_Simple")
# Shapefile of amazon legal area can also be obtained at http://www.dpi.inpe.br/Ambdata/unidades_administrativas.php
shapeINPE <- read_sf(dsn = "./geo_data/AmLeg_LLwgs84", layer = "amazlegal")
```
```{r}
legal_amazon <- as.data.frame(shape$geometry[[1]][[1]])
names(legal_amazon) <- c("longitude", "latitude")
legal_amazon_INPE <- as.data.frame(shapeINPE$geometry[[1]][[1]][[1]])
names(legal_amazon_INPE) <- c("longitude", "latitude")
legal_amazon %>% ggplot(aes(x = longitude, y = latitude)) +
geom_polygon()
legal_amazon_INPE %>% ggplot(aes(x = longitude, y = latitude)) +
geom_polygon()
# To see if a point is inside a polygon
# point.in.polygon(-60, -5, shape$geometry[[1]][[1]][,1], shape$geometry[[1]][[1]][,2])
```
How much of the dataset is inside the legal amazon area (from january to september)
```{r}
dfs <- list(data15, data16, data17, data18, data19)
scores <- c()
for (i in 1:5) {
scores[i] <- paste("Observations detected inside the legal amazon in", unique(year(dfs[[i]]$datahora)), "are",
round(sum(point.in.polygon(dfs[[i]]$longitude, dfs[[i]]$latitude, shape$geometry[[1]][[1]][,1], shape$geometry[[1]][[1]][,2])) / nrow(dfs[[i]]), 2))
}
writeLines(scores)
rm(dfs)
rm(scores)
```
```{r}
# Only for 2019
data %>%
select(estado) %>%
group_by(estado) %>%
summarise(n = n()) %>%
right_join(map_df, by = 'estado') %>%
ggplot(aes(x = longitude, y = latitude, group = codigo)) +
geom_polygon() +
geom_point(inherit.aes = F, data = data19, aes(x = longitude, y = latitude, col = bioma), size = 0.1, shape = 21, alpha = 1/20) +
scale_color_manual(values = cbp1) +
geom_path(color = 'white', size = 0.1) +
geom_path(inherit.aes = F, data = legal_amazon_INPE, mapping = aes(x = longitude, y = latitude), color = "yellow") +
geom_dl(aes(label = "Amazônia Legal"), method = list(dl.combine("last.points"), cex = 0.6, hjust = 3, vjust = -14)) +
guides(colour = guide_legend(title = "Bioma", override.aes = list(alpha = 1))) +
# facet_wrap(~ano) +
labs(title = 'Distribuição espacial de focos de incêndios', subtitle = "Entre Janeiro e Setembro de 2019", caption="Fonte: Elaborado pelo Autor" ) +
theme(plot.title=element_text(size=9,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=7,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title = element_blank(),
axis.text = element_blank(),
axis.ticks = element_blank(),
legend.position=c(0.1, 0.45),
legend.justification=c(0, 1),
legend.key.width= unit(0.6, "lines"),
legend.key.height = unit(0.6, "lines"),
legend.title = element_text(size = 7),
legend.text = element_text(size = 5),
plot.caption = element_text(size = 6))
# ggsave(paste0(path_ch4, "amazon2019.png"), dpi = "retina", width = 10, height = 8, units = "cm")
rm(data19)
```
It is worth noting not only the distribution of point occurrences in 2019 but also through all years involved in the analysis process
```{r}
# For all years
data %>%
select(estado) %>%
group_by(estado) %>%
summarise(n = n()) %>%
right_join(map_df, by = 'estado') %>%
ggplot(aes(x = longitude, y = latitude, group = codigo)) +
geom_polygon() +
geom_point(inherit.aes = F, data = data_partial, aes(x = longitude, y = latitude, col = bioma), size = 0.01, shape = 21, alpha = 1/50) +
scale_color_manual(values = cbp1) +
geom_path(color = 'white', size = 0.1) +
geom_path(inherit.aes = F, data = legal_amazon_INPE, mapping = aes(x = longitude, y = latitude), color = "yellow") +
geom_dl(aes(label = "Amazônia Legal"), method = list(dl.combine("last.points"), cex = 0.6, hjust = 3, vjust = -14)) +
guides(colour = guide_legend(title = "Bioma", override.aes = list(alpha = 1))) +
facet_wrap(~ano) +
labs(title = 'Distribuição espacial de focos de incêndios', subtitle = "Entre Janeiro e Setembro", caption="Fonte: Elaborado pelo Autor" ) +
theme(plot.title=element_text(size=9,
face="bold",
family="Arial",
hjust=0.5,
lineheight=1.2), # title
plot.subtitle=element_text(size=7,
family="Arial",
face="bold",
hjust=0.5), # subtitle
axis.title = element_blank(),
axis.text = element_blank(),
axis.ticks = element_blank(),
legend.position=c(0.7, 0.5),
legend.justification=c(0, 1),
legend.key.width= unit(0.6, "lines"),
legend.key.height = unit(0.6, "lines"),
legend.title = element_text(size = 7),
legend.text = element_text(size = 5),
plot.caption = element_text(size = 6))
# ggsave(paste0(path_ch4, "amazon-allYears.png"), dpi = "retina", width = 10, height = 8, units = "cm")
rm(data_partial)
```
What about cumulative sums?
```{r}
accum_fires_inpe <- data %>%
mutate(data = date(datahora),
ano = year(datahora)) %>%
select(ano, data) %>%
group_by(ano, data) %>%
summarise(occurrence = n()) %>%
mutate(cum_sum = cumsum(occurrence),
dia_ano = yday(data))
accum_fires_inpe %>% ggplot(aes(x = data, y = cum_sum)) +
geom_line() +
scale_x_date(date_breaks = "4 months", date_labels = "%b/%y")
```
```{r}
splitted <- accum_fires_inpe %>% select(ano, dia_ano, cum_sum)
# Using `do` function to replace NAs values with the last records value
tidy <- splitted %>% spread(key = ano, value = cum_sum) %>% do(zoo::na.locf(.))
tidy$`2019`[273:nrow(tidy)] <- NA
tidy <- tidy %>%