-
Notifications
You must be signed in to change notification settings - Fork 1
/
125_Calculate_trends_leftadjusted.Rmd
1235 lines (876 loc) · 31.9 KB
/
125_Calculate_trends_leftadjusted.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: "125 Calculate time trends using package leftadjusted"
output: html_document
---
This procedure uses data from script 109
**Basis has now been added to all relevant "series" files, the plot result files, and functions (scr. 125...functions + 402..functions**
**This was done for results in folder 07, see APPENDIX 2 for modificating and combining files at the bottom of this script**
**HOWEVER: The codes in this file should be made so series also are selected using 'Basis'**
## Install leftcensored
- And make sure it's the right version! See comment below.
```{r}
remove.packages("leftcensored")
devtools::install_github("DagHjermann/leftcensored", upgrade = "never", force = TRUE, )
# Check model syntax
cat(leftcensored:::get_jags_model_code())
# LINE 10 SHOULD have 'min(max(...', as this:
# prob[j] <- min(max(pnorm(cut[j], mu[n+j], tau), 0.01),0.99)
source("https://raw.githubusercontent.com/DagHjermann/leftcensored/main/R/jags_model_code.R")
# Check model syntax
cat(leftcensored:::get_jags_model_code())
cat(get_jags_model_code())
# dir("../../../../../opt/R")
# dir("../../../../../opt/R/leftcensored")
# dir("../../../../../opt/R/leftcensored/R/leftcensored")
# file.remove
```
## 0a. Constants (update each year)
```{r}
last_year <- 2021
# Not used:
series_lastyear <- 2018 # The series must last at least until this year
# Series without data in any of the three last years will be excluded
#
# Folder for saving input files and results
#
# folder_5: full new run of all series, JAGS model including "prob[j] <- min(max(..." (see above)
# folder_6: time trends of length-adjusted series, only HG so far
folder_results <- "Data/125_results_2021_05" # Basis WW, all
# folder_results <- "Data/125_results_2021_06" # Basis WWa (length-adjusted), HG only
# folder_results <- "Data/125_results_2021_07" # combining folder 05 and 06, see APPENDIX 2 below
folder_input <- paste0(folder_results, "_input")
# folder_results <- paste0("Data/125_results_", last_year)
#
if (!dir.exists(folder_results)){
dir.create(folder_results)
}
if (!dir.exists(folder_input)){
dir.create(folder_input)
}
```
### Type of value (wet-weight, wet-weight length-adjusted, etc.)
```{r}
if (folder_input == "Data/125_results_2021_05_input"){
# For runs up until folder_5:
value_column <- "VALUE_WW"
} else if (folder_input == "Data/125_results_2021_05_input"){
# For folder_6:
value_column <- "VALUE_WWa"
} else {
stop("Latest results are folder 5 or 6 - select one of these")
}
message("Selected value column: ", value_column)
```
## 1. Libraries and functions
### The rest
```{r, results='hide', message=FALSE, warning=FALSE}
# install.packages("lubridate")
# General purpose
library(dplyr)
library(tidyr)
library(purrr)
library(mgcv) # mgcv_1.8-39
# install.packages("mgcv") # for mgcv_1.8-40, b
# packageVersion("mgcv")
packageVersion("mgcv")
library(ggplot2)
# Specific for the analysis
library(rjags)
library(runjags)
library(leftcensored)
# For parallel computing
if (!"doParallel" %in% installed.packages())
install.packages("doParallel")
library(doParallel)
# Load functions defined in other scripts
source("125_Calculate_trends_leftadjusted_functions.R")
source("110_Medians_and_PROREF_functions.R") # for homogenize_series
source("002_Utility_functions.R")
```
## 2. Read data
### a. Main data
Read and reformat the most recent data (by default)
```{r, collapse=TRUE}
# If we have NOT length-adjusted the last year's data:
# filepattern <- "105_data_with_uncertainty_" # entire file name except date and extension
# Normally, if we have length-adjusted the last year's data:
filepattern <- "109_adjusted_data_" # entire file name except date and extension
filenumber <- 1 # filenumber = 1 means "read the newest file"
files <- dir("Data", pattern = filepattern) %>% rev()
dat_all <- read_rds_file("Data",
files,
filenumber = filenumber, # "1" gets the newest file
get_date = FALSE, time_since_modified = TRUE)
# dat_all <- data_list$data
# The file_date text will be used in part 10, when we save the resulting file
# cat("File date text:", data_list$file_date, "\n")
```
### b. Homogenize time series
Change STATION_CODE, in order to make a single time series for data with different STATION_CODE that in practice should count as the same station
* Fixing station 227G1 and 227G2 (shall count as 227G)
* Fixing station 36A and 36A1 (36A1 shall count as 36A)
Also combines PARAM = VDSI and PARAM = Intersex to PARAM = "VDSI/Intersex" for station 71G
```{r}
dat_all <- homogenize_series(dat_all)
```
### c. Prepare data
- **Important**: set 'value_column' (WW or WWa)
- NOTE: part c-d1 (d1 at least) are slow and can be skipped by going to d2 and read the saved data
```{r}
# These warnings can be ignored:
# Warning: Unknown or uninitialised column: `threshold`.
# Warning: Unknown or uninitialised column: `uncensored`.
dat_all_isotopes <- dat_all %>%
filter(grepl("Delta", PARAM))
dat_all_with_zeros <- dat_all %>%
filter(!grepl("Delta", PARAM)) %>%
group_by(PARAM, STATION_CODE, TISSUE_NAME, LATIN_NAME) %>%
mutate(N_zeros = sum(VALUE_WW <= 0)) %>%
filter(N_zeros > 0)
dat_all_sans_zeros <- dat_all %>%
filter(!grepl("Delta", PARAM)) %>%
group_by(PARAM, STATION_CODE, TISSUE_NAME, LATIN_NAME) %>%
mutate(N_zeros = sum(VALUE_WW <= 0)) %>%
filter(N_zeros == 0)
dat_all_prep1_isotopes <- lc_prepare(dat_all_isotopes,
x = "MYEAR",
y = value_column,
censored = "FLAG1",
log = FALSE,
keep_original_columns = TRUE)
dat_all_prep1_with_zeros <- lc_prepare(dat_all_with_zeros,
x = "MYEAR",
y = value_column,
censored = "FLAG1",
log = TRUE, const = 1,
keep_original_columns = TRUE)
dat_all_prep1_sans_zeros <- lc_prepare(dat_all_sans_zeros,
x = "MYEAR",
y = value_column,
censored = "FLAG1",
log = TRUE,
keep_original_columns = TRUE)
dat_all_prep1 <- bind_rows(
dat_all_prep1_isotopes,
dat_all_prep1_with_zeros,
dat_all_prep1_sans_zeros
)
# For folder_results = 125_results_2021_06:
# dat_all_prep1 <- dat_all_prep1_sans_zeros
# Test
# lc_plot(dat_all_prep1 %>% filter(PARAM %in% "CB153" & STATION_CODE %in% "15A"))
# lc_plot(dat_all_prep1 %>% filter(PARAM %in% "SN" & STATION_CODE %in% "80B"))
```
### c2. HG ONLY
**NOTE**
```{r}
# dat_all_prep1 <- dat_all_prep1 %>%
# filter(PARAM %in% "HG")
```
### c3. Sum parameters
*Not used*
Idea was to use the leftcensored method for sums, setting trends based on agreement between trends for "without LOQ" and "with LOQ" data.
- E.g. set downward trend if bothtremds of "without LOQ" and "with LOQ" data have a downwards trend, otherwise set trend unknown
```{r}
if (FALSE){
source("101_Combine_with_legacy_data_functions.R") # list 'sum_parameters'
names(sum_parameters)
# sum_parameters[["DDTEP"]]
# sum_parameters[["CB_S7"]]
sum_parameters[["BDE6S"]]
year1 <- 2011
sumpar <- "BDE6S"
df_without_loq <- dat_all_prep1 %>%
filter(PARAM %in% sum_parameters[[sumpar]] & MYEAR >= year1) %>%
mutate(
VALUE_WW = case_when(
is.na(FLAG1) ~ VALUE_WW,
!is.na(FLAG1) ~ 0)
) %>%
group_by(STATION_CODE, TISSUE_NAME, LATIN_NAME, x) %>%
summarize(
VALUE_WW = sum(VALUE_WW)
) %>%
ungroup() %>%
mutate(
PARAM = paste0(sumpar, "_without_LOQ"),
FLAG1 = as.character(NA)
)
df_with_loq <- dat_all_prep1 %>%
filter(PARAM %in% sum_parameters[[sumpar]] & MYEAR >= year1) %>%
group_by(STATION_CODE, TISSUE_NAME, LATIN_NAME, x) %>%
summarize(
VALUE_WW = sum(VALUE_WW)
) %>%
ungroup() %>%
mutate(
PARAM = paste0(sumpar, "_with_LOQ"),
FLAG1 = as.character(NA)
)
df_with_and_without_loq <- bind_rows(df_without_loq, df_with_loq)
}
# dat_all_prep1 <- dat_all_prep1 %>%
# filter(PARAM %in% "HG")
```
### d1. Add flags for rule 1 and rule 2 (columns Rule1 and Rule2)
```{r}
if (FALSE){
# Testing
test1 <- dat_all_prep1 %>% filter(PARAM %in% "SN" & STATION_CODE %in% "80B")
test2 <- leftcensored:::lc_flag1(test1)
test3 <- leftcensored:::lc_flag2(test2)
}
#
# Create 'dat_all_prep3'
#
# File name for saving data
fn_dat_all <- paste0(folder_input, "/125_dat_all_prep3.rds")
if (!file.exists(fn_dat_all)){
# If file doesn't exist, we create it and store it
lc_flag1_seriesno <- function(i, seriesdata, data){
data %>%
filter(PARAM %in% seriesdata$PARAM[i],
STATION_CODE %in% seriesdata$STATION_CODE[i],
LATIN_NAME %in% seriesdata$LATIN_NAME[i],
TISSUE_NAME %in% seriesdata$TISSUE_NAME[i]) %>%
leftcensored:::lc_flag1(show_result = FALSE)
}
lc_flag2_seriesno <- function(i, seriesdata, data){
data %>%
filter(PARAM %in% seriesdata$PARAM[i],
STATION_CODE %in% seriesdata$STATION_CODE[i],
LATIN_NAME %in% seriesdata$LATIN_NAME[i],
TISSUE_NAME %in% seriesdata$TISSUE_NAME[i]) %>%
leftcensored:::lc_flag2(show_result = FALSE)
}
# Only series existing in the last year
dat_series_for_flag <- dat_all_prep1 %>%
filter(MYEAR == last_year) %>%
distinct(PARAM, STATION_CODE, LATIN_NAME, TISSUE_NAME) %>%
as.data.frame()
dat_series_for_flag$Rowno <- 1:nrow(dat_series_for_flag) # only for interactive checking
if (FALSE){
# Testing
test2 <- lc_flag1_seriesno(1537, dat_series_for_flag, dat_all_prep1)
test3 <- lc_flag2_seriesno(1537, dat_series_for_flag, dat_all_prep1)
xtabs(~Rule2 + MYEAR, test3)
}
# SLOW! 9 minutes for flag1, 4 minutes fpr flag2 - but it works, at least
dat_all_prep2 <- map_dfr(
1:nrow(dat_series_for_flag), lc_flag1_seriesno,
seriesdata = dat_series_for_flag,
data = dat_all_prep1)
dat_all_prep3 <- map_dfr(
1:nrow(dat_series_for_flag), lc_flag2_seriesno,
seriesdata = dat_series_for_flag,
data = dat_all_prep2)
# test4 <- dat_all_prep3 %>% filter(PARAM %in% "SN" & STATION_CODE %in% "80B")
saveRDS(dat_all_prep3, fn_dat_all)
} else {
# If file does exist, we just read it
dat_all_prep3 <- readRDS(fn_dat_all)
}
```
### e. Make dat_series
- One line per time series
```{r}
# version of max() and min() that tolerates being given an x with length zero without warning
max_warningless <- function(x)
ifelse(length(x)==0, NA, max(x))
min_warningless <- function(x)
ifelse(length(x)==0, NA, min(x))
# TEST
# sel <- c(F,F,T,T,F)
# max_warningless((1:5)[sel])
# sel <- rep(FALSE,5)
# max_warningless((1:5)[sel])
dat_series_1 <- dat_all_prep3 %>%
group_by(PARAM, STATION_CODE, TISSUE_NAME, LATIN_NAME, x) %>%
summarise(
N = n(),
N_over_LOQ = sum(uncensored == 1),
P_over_LOQ = N_over_LOQ/N,
Rule1 = first(Rule1),
Rule2 = first(Rule2),
.groups = "drop") %>%
group_by(PARAM, STATION_CODE, TISSUE_NAME, LATIN_NAME) %>%
summarise(
First_year_overall = min(x),
Last_year_overall = max(x),
First_year = min_warningless(x[Rule1 & Rule2]),
Last_year = max_warningless(x[Rule1 & Rule2]),
N_years = length(unique(x[Rule1 & Rule2])),
N_years_10yr = length(unique(x[Rule1 & Rule2 & x >= (last_year-10)])),
Years_over_LOQ = sum(Rule1 & Rule2 & N_over_LOQ > 0),
Last_year_over_LOQ = max_warningless(x[Rule1 & Rule2 & N_over_LOQ > 0]),
.groups = "drop")
if (FALSE){
dat_series_1 %>% filter(PARAM %in% "CB28") %>% View()
dat_all_prep3 %>% filter(PARAM %in% "CB28" & STATION_CODE %in% "98A2") %>% lc_plot()
}
dat_series <- dat_series_1 %>%
filter(Last_year_overall >= last_year) %>%
mutate(
Trend_model = case_when(
Years_over_LOQ %in% 0:1 ~ "No model",
Years_over_LOQ %in% 2 & N_years %in% 2 ~ "No model",
Years_over_LOQ %in% 2:4 & N_years >= 3 ~ "Mean",
Years_over_LOQ %in% 5:6 ~ "Linear",
Years_over_LOQ %in% 7:9 ~ "Smooth, k_max=3",
Years_over_LOQ %in% 10:14 ~ "Smooth, k_max=4",
Years_over_LOQ >= 15 ~ "Smooth, k_max=5"),
Trend_model = factor(
Trend_model,
levels = c("No model", "Mean", "Linear",
"Smooth, k_max=3", "Smooth, k_max=4", "Smooth, k_max=5")),
k_max = case_when(
Trend_model %in% "No model" ~ as.numeric(NA),
Trend_model %in% "Mean" ~ 1,
Trend_model %in% "Linear" ~ 2,
Trend_model %in% "Smooth, k_max=3" ~ 3,
Trend_model %in% "Smooth, k_max=4" ~ 4,
Trend_model %in% "Smooth, k_max=5" ~ 5)
)
nrow(dat_series)
table(dat_series$Trend_model)
# xtabs(~Substance.Group + Trend_model, dat_series)
```
### f. Statistics per substance
- Selected substances
```{r}
dat_series %>%
filter(PARAM %in% c("HG", "CD", "CB118", "BAP", "PFOS", "PFOA")) %>%
count(PARAM, Trend_model) %>%
pivot_wider(PARAM, names_from = Trend_model, values_from = n, values_fill = 0,
names_sort= TRUE)
```
### g. Last_year_over_LOQ
```{r}
# Last_year_over_LOQ vs Last_year
xtabs(~Last_year + Last_year_over_LOQ, dat_series)
```
### h. Data series to run
#### For first-time run
```{r}
dat_series_trend <- dat_series %>%
filter(!Trend_model %in% "No model") #%>%
# filter(PARAM %in% "HG")
# 'series_no' decides name of result file that will be (over)written
dat_series_trend$series_no <- 1:nrow(dat_series_trend)
```
#### For finishing run that has started, but was interrupted
See APPENDIX 1 below
### i. Save series (and read saved files)
```{r}
saveRDS(dat_series,
paste0(folder_input, "/125_dat_series.rds"))
saveRDS(dat_series_trend,
paste0(folder_input, "/125_dat_series_trend.rds"))
if (FALSE){
# Read back
dat_all_prep3 <- readRDS(paste0(folder_input, "/125_dat_all_prep3.rds"))
dat_series <- readRDS(paste0(folder_input, "/125_dat_series.rds"))
dat_series_trend <- readRDS(paste0(folder_input, "/125_dat_series_trend.rds"))
# If series lacking:
# dat_series_trend <- readRDS(paste0(folder_input, "/125_dat_series_trend_lacking.rds"))
}
```
### x. add "new" "No model" series to 'dat_series' of results folder 7
```{r}
if (FALSE){
# ONE TIME ONLY:
# add "new" "No model" series to 'dat_series' of results folder 7
folder_input_5 <- "Data/125_results_2021_05_input"
folder_input_7 <- "Data/125_results_2021_07_input"
dat_series_5 <- readRDS(paste0(folder_input_5, "/125_dat_series.rds")) %>%
mutate(Basis = "WW")
dat_series_7 <- readRDS(paste0(folder_input_7, "/125_dat_series.rds"))
# Backup of old version:
# file.copy(
# paste0(folder_input_7, "/125_dat_series.rds"),
# paste0(folder_input_7, "/125_dat_series_OLD.rds")
# )
dat_series_to_add <- dat_series_5 %>%
anti_join(dat_series_7, by = c("PARAM", "STATION_CODE", "TISSUE_NAME", "LATIN_NAME", "Basis"))
nrow(dat_series_to_add)
# get and add Substance.Group
lookup_substgroup <- dat_series_7 %>% distinct(PARAM, Substance.Group)
dat_series_to_add <- dat_series_to_add %>%
left_join(lookup_substgroup)
# get and add Substance.Group
nrow(dat_series_5)
nrow(dat_series_to_add)
nrow(dat_series_7) # 2393
nrow(dat_series_to_add) + nrow(dat_series_7) # 3022 - larger than 'dat_series_5', as 'dat_series_7' also contains WWa for HG
dat_series_7_new <- bind_rows(dat_series_7, dat_series_to_add)
nrow(dat_series_7)
nrow(dat_series_7_new)
# Overwrite old file:
# ONE TIME ONLY!
# saveRDS(dat_series_7_new,
# paste0(folder_input_7, "/125_dat_series.rds"))
}
```
## 3. Test
### Single series
```{r}
# param <- "AG"
# station_code <- "30A"
param <- "HG"
station_code <- "30B"
param <- "PB"
station_code <- "02B"
param <- "MCCP eksl. LOQ"
station_code <- "71B"
param <- "CB101"
station_code <- "30B"
#
# Example 1
#
# MCMC stops - Compilation error on line 22. Unknown variable zero
param <- "HG"
station_code <- "36A"
# Fairly fast
param <- "CB52"
station_code <- "65A"
data_prep <- dat_all_prep3 %>%
filter(PARAM == param & STATION_CODE == station_code)
i <- with(dat_series_trend, which(PARAM == param & STATION_CODE == station_code))
i
dat_series_trend[i,]
lc_plot(data_prep)
last_year_over_LOQ <- dat_series %>%
filter(PARAM == param & STATION_CODE == station_code) %>% # View
pull(Last_year_over_LOQ)
k_max <- dat_series %>%
filter(PARAM == param & STATION_CODE == station_code) %>%
pull(k_max)
k_values <- 1:k_max
# k_values <- 2:k_max
raftery <- FALSE
if (FALSE){
# Test for a single 'k' value
# debugonce(leftcensored::lc_fixedsplines_tp)
check <- leftcensored::lc_fixedsplines_tp(
data = data_prep,
k = 2,
normalize = FALSE, raftery = raftery, measurement_error = "Uncertainty",
predict_x = seq(min(data_prep$x), max(data_prep$x), by = 0.25),
reference_x = last_year_over_LOQ, set_last_equal_x = last_year_over_LOQ)
View(check$plot_data)
}
if (FALSE){
# Test for all 'k' values ('k_values')
results_all <- purrr::map(
k_values,
~leftcensored::lc_fixedsplines_tp(
data = data_prep,
k = .x,
normalize = FALSE, raftery = raftery, measurement_error = "Uncertainty",
predict_x = seq(min(data_prep$x), max(data_prep$x)),
reference_x = last_year_over_LOQ, set_last_equal_x = last_year_over_LOQ)
)
names(results_all) <- k_values
str(results_all, 1)
str(results_all[[1]], 1)
str(results_all[[1]]$deviance, 1)
dev <- map(results_all, "deviance", .id = "k") %>% map_dbl(sum)
ddev <- -diff(dev)
p_values <- 1-pchisq(ddev, 1)
data.frame(k = names(dev), dev=dev, ddev = c(NA,ddev), p = c(NA,p_values))
}
# lc_plot(data_prep)
lc_plot(data_prep, results = results_all, facet = "wrap")
```
### get_splines_results_seriesno
```{r, results='hide', message=FALSE, warning=FALSE}
# Fairly fast
param <- "CB52"
# param <- "HG"
station_code <- "65A"
param <- "HG"
station_code <- "43B2"
get_seriesno(param, station_code, data_series = dat_series_trend)
# debugonce(get_splines_results_seriesno)
get_splines_results_seriesno_s(11,
data = dat_all_prep3,
data_series = dat_series_trend,
foldername = folder_results, # test site
raftery = FALSE)
```
### Check result
```{r}
i <- 409
fn <- sprintf("trend_%04i.rda", i)
check <- readRDS(paste0(folder_results, "/", fn))
str(check, 1)
get_pointdata <- function(seriesno, data, data_series){
subset(data,
PARAM %in% data_series$PARAM[seriesno] &
STATION_CODE %in% data_series$STATION_CODE[seriesno] &
TISSUE_NAME %in% data_series$TISSUE_NAME[seriesno] &
LATIN_NAME %in% data_series$LATIN_NAME[seriesno])
}
# debugonce(get_pointdata)
df_points <- get_pointdata(i, data = dat_all_prep3, data_series = dat_series_trend)
ggplot(check$plot_data, aes(x, y)) +
geom_ribbon(aes(ymin = y_q2.5, ymax = y_q97.5), fill = "lightblue") +
geom_point(data = df_points) +
geom_point(data = df_points, aes(y = threshold), shape = 6) +
geom_line()
```
## 4. Analysing time trends and saving results
- We save results as R files as we go, in case something goes wrong
### Run analyses using doParallel
```{r, results='hide'}
if (FALSE){
# if (T){
# Do only once
future::availableCores()
# 4 / 16 / 64
cl <- makeCluster(4)
registerDoParallel(cores = 4)
# stopCluster(cl)
}
#
# Parallel
#
# NEW: seriesno = column in the 'dat_series' data
series_no <- dat_series_trend$series_no
# series_no <- c(6,7,8)
range(series_no)
length(series_no) # 1507
# Note folder name!
check_contents <- dir(folder_results)
if (length(check_contents) > 0){
warning("You are writing to a folder which already contains ", length(check_contents), " files. THESE MAY BE OVERWRITTEN.")
}
# Run analyses
# For following the number of files from the terminal, use
# ls -lrt
# find -newermt '2 hours ago' | wc -l
# (1) lists files with the newest file last, the second counts number of files younger than 2 hour
t0 <- Sys.time()
result <- foreach(i = series_no,
.export = c("dat_all_prep3", "dat_series_trend")) %dopar%
get_splines_results_seriesno_s(i,
dat_all_prep3, dat_series_trend, foldername = folder_results,
raftery = TRUE)
t1 <- Sys.time()
t1-t0
# 30 min for 100 on 16 cores
# 4.8 hours for 1960 on 64 cores
```
### Run analyses without doParallel
```{r}
for (i in c(15:17)){
result <- get_splines_results_seriesno_s(i,
data = dat_all_prep3,
data_series = dat_series_trend,
foldername = folder_results, # test site
raftery = TRUE)
}
# i = 14: HG 71B
# no non-missing arguments to max; returning -Inf>
# dat_series_trend[14,]
```
## 5. Check results
### Read and quick check of files
```{r}
# folder_results <- "Data/125_results_2021_06"
# File names and numbers
fns <- dir(folder_results, full.names = TRUE) %>% sort()
length(fns)
fileinfo_no <- substr(fns, nchar(folder_results) + 8, nchar(folder_results) + 11)
names(fns) <- fileinfo_no
# Read file content
result_list <-lapply(fns, readRDS)
# Check length of each result
result_length <- map_int(result_list, length)
table(result_length)
```
### Extract data frame for "successes"
- Meaning that fitting worked (DIC exists)
```{r}
jags_finished <- map_lgl(result_list, ~!is.null(.x$k_values))
ok <- map_lgl(result_list, ~!is.null(.x$DIC))
plotno <- map_dbl(result_list, "seriesno")
PARAM <- map_chr(result_list, "PARAM")
STATION_CODE <- map_chr(result_list, "STATION_CODE")
TISSUE_NAME <- map_chr(result_list, "TISSUE_NAME")
LATIN_NAME <- map_chr(result_list, "LATIN_NAME")
k_max <- map_chr(result_list, "k_max")
k_sel <- NA
k_sel[ok] <- map_int(result_list[ok], "k_sel")
dat_success <- data.frame(ok, jags_finished, plotno, PARAM, STATION_CODE, TISSUE_NAME, LATIN_NAME, k_max, k_sel)
xtabs(~jags_finished + ok, dat_success)
xtabs(~k_max + ok, dat_success)
```
### Plot a single series
```{r}
tsplot_seriesno(58, folder = folder_results)
tsplot_seriesno(58, folder = folder_results)
```
### Check one group of compounds
```{r}
pno <- dat_success %>%
filter(PARAM == "MCCP eksl. LOQ" & ok) %>%
pull(plotno)
df_modelfit <- map_dfr(pno, extract_modelfit_data, folder = folder_results)
df_rawdata <- map_dfr(pno, extract_raw_data)
ggplot(df_modelfit, aes(x, y)) +
geom_ribbon(aes(ymin = y_q2.5, ymax = y_q97.5), fill = "lightblue") +
geom_point(data = df_rawdata %>% filter(!is.na(y))) +
geom_point(data = df_rawdata %>% filter(!is.na(threshold)), aes(y = threshold), shape = 6) +
geom_line() +
facet_wrap(vars(STATION_CODE), scales = "free_y")
```
## 6. Check Rule1 and Rule2
- Rule 1. Time series should be truncated from the left until Nplus/N >= 0.5
- Rule 2. If a linear/smooth trend is fitted, the first year must be non-censored
- Forgot to think of these when the estimations were ran
```{r}
df_rule1 <- dat_all_prep3 %>%
group_by(PARAM, STATION_CODE, TISSUE_NAME, LATIN_NAME) %>%
summarise(Rule1_first = first(Rule1))
table(df_rule1$Rule1_first)
df_rule2 <- dat_all_prep3 %>%
group_by(PARAM, STATION_CODE, TISSUE_NAME, LATIN_NAME) %>%
summarise(Rule2_first = first(Rule2))
table(df_rule2$Rule2_first)
dat_success <- dat_series_trend %>%
left_join(df_rule1) %>%
left_join(df_rule2)
table(dat_success$Rule1_first) # 6 - should be run again
table(dat_success$Rule2_first) # 20 - should be run again
```
## APPENDIX 1
#### For finishing run that has started, but was interrupted
```{r}
if (FALSE){
# FIRST, RE-MAKE 'dat_series_trend'
# THEN:
# File names and numbers
fns <- dir(folder_results, full.names = TRUE) %>% sort()
length(fns)
fileinfo_no <- substr(fns, nchar(folder_results) + 8, nchar(folder_results) + 11)
names(fns) <- fileinfo_no
# Read file content
result_list <-lapply(fns, readRDS)
result_length <- map_int(result_list, length)
table(result_length)
# 5 7 20
# 38 405 1233
if (FALSE){
# Examples
i <- which(result_length == 5)[1]
str(result_list[[i]], 1)
i <- which(result_length == 7)[1]
str(result_list[[i]], 1)
i <- which(result_length == 20)[1]
str(result_list[[i]], 1)
}
serno_incomplete <- fileinfo_no[result_length == 5] %>% as.numeric()
serno_lacking <- which(!dat_series_trend$series_no %in% as.numeric(fileinfo_no))
serno_torun <- c(serno_incomplete, serno_lacking)
# Update dat_series_trend
nrow(dat_series_trend)
dat_series_trend <- dat_series_trend[serno_torun,]
nrow(dat_series_trend)
# DON'T save dat_series_trend (i.e.. don't run 2.i)
# GO DIRECTLY TO 4
}
```
## APPENDIX 2 - Combine WW and WWa runs
- folder 07
- Ths will be used for plotting etc.
- combine files from folders 05 (WW) and 06 (WWa)
- As we stupidly didn't included 'Basis' anywhere, some changes needs to be done
```{r}
# folder_results <- "Data/125_results_2021_05" # WW (wet-weight), all parameters
# folder_results <- "Data/125_results_2021_06" # WWa (wet-weight length-adjusted), HG only
#
# 1. Made folder 'Data/125_results_2021_07'
#
#
# 2. Set these