From 24f43049cbc5767e4ff33f09be5047b062e86f3f Mon Sep 17 00:00:00 2001 From: RCJung Date: Fri, 6 Dec 2024 16:09:19 +0100 Subject: [PATCH] tscount 1 --- Rnotebook-tscounts.html | 2025 +++++++++++++++++ Rnotebook-tscounts.md | 460 ++++ .../unnamed-chunk-12-1.png | Bin 0 -> 4401 bytes .../unnamed-chunk-12-2.png | Bin 0 -> 6120 bytes .../unnamed-chunk-16-1.png | Bin 0 -> 4333 bytes .../unnamed-chunk-16-2.png | Bin 0 -> 6144 bytes .../unnamed-chunk-17-1.png | Bin 0 -> 5075 bytes .../unnamed-chunk-17-2.png | Bin 0 -> 4506 bytes .../unnamed-chunk-17-3.png | Bin 0 -> 4495 bytes .../unnamed-chunk-19-1.png | Bin 0 -> 4518 bytes .../unnamed-chunk-19-2.png | Bin 0 -> 6357 bytes .../unnamed-chunk-21-1.png | Bin 0 -> 4415 bytes .../unnamed-chunk-21-2.png | Bin 0 -> 6107 bytes .../unnamed-chunk-6-1.png | Bin 0 -> 4231 bytes .../unnamed-chunk-6-2.png | Bin 0 -> 6024 bytes gitlab-readme.txt | 13 + 16 files changed, 2498 insertions(+) create mode 100644 Rnotebook-tscounts.html create mode 100644 Rnotebook-tscounts.md create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-12-1.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-12-2.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-16-1.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-16-2.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-1.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-2.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-3.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-19-1.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-19-2.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-21-1.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-21-2.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-6-1.png create mode 100644 Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-6-2.png create mode 100644 gitlab-readme.txt diff --git a/Rnotebook-tscounts.html b/Rnotebook-tscounts.html new file mode 100644 index 0000000..4b53097 --- /dev/null +++ b/Rnotebook-tscounts.html @@ -0,0 +1,2025 @@ + + + + + + + + + + + + + +R Notebook tscounts + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
+ + + + + + + +

THIS VERSION: 6.12.2024

+ +

In this document I try to fit the UK strikes data and the rig count +data that we propose in the JSS paper to be analysed with the \(coconots\) package with the \(tscount\) package.

+

I start with the UK strikes data.

+

Then I present an analysis of the rig counts in the time span 2014 to +2020. The tscount package does a remarkably good job in capturing the +dependence structure, but struggles with the slight underdispersion in +the data. Values of the scoring rules are considerably larger (=worse) +as those from the GPAR2 model.

+

Finally, I then present results for rig count in the time span 2017 +to 2024. Here I find no specification of the models in tscount, that can +capture the dynamics in the data satisfactorily.

+
+

UK Strikes Data

+
data_uks <- read_excel("data_strikes.xlsx")
+

I start with the UK strikes data (public sector). In \(coconots\), we propose a GPAR2 model with +harmonics and a linear trend (using a log-link).

+

I generate a holdout sample of 3, and prepare the regressors +first:

+
hos <- 3
+
+X <- data_uks$public
+n <- length(X) - hos
+
+sin_x <- sin(2*pi*1:n/12)
+cos_x <- cos(2*pi*1:n/12)
+trend <- (1:n - n/2) / n
+const <- rep(1, n)
+xregtsc <- cbind(const, trend, cos_x, sin_x)
+

Then we fit an INGARCH(1,1) type of model with a log link and +negative binomial conditional distributional (Poisson led to a slightly +u-shaped PIT histogram) assumption:

+
uks_tsc_fit <- tsglm(data_uks$public[1:n], model = list(past_obs = c(1),past_mean=1), link = "log", distr = "nbinom", xreg = xregtsc)
+

The standard regression output:

+
summary(uks_tsc_fit)
+
## 
+## Call:
+## tsglm(ts = data_uks$public[1:n], model = list(past_obs = c(1), 
+##     past_mean = 1), xreg = xregtsc, link = "log", distr = "nbinom")
+## 
+## Coefficients:
+##              Estimate  Std.Error  CI(lower)  CI(upper)
+## (Intercept)    1.2547     0.8288    -0.3697   2.879134
+## beta_1         0.3990     0.0670     0.2677   0.530197
+## alpha_1        0.2153     0.1382    -0.0556   0.486154
+## const         -0.4992     0.6884    -1.8485   0.850020
+## trend         -0.3439     0.1196    -0.5783  -0.109619
+## cos_x         -0.0609     0.0309    -0.1214  -0.000343
+## sin_x          0.0165     0.0306    -0.0434   0.076477
+## sigmasq        0.0547         NA         NA         NA
+## Standard errors and confidence intervals (level =  95 %) obtained
+## by normal approximation.
+## 
+## Link function: log 
+## Distribution family: nbinom (with overdispersion coefficient 'sigmasq') 
+## Number of coefficients: 8 
+## Log-likelihood: -729.3632 
+## AIC: 1474.726 
+## BIC: 1503.974 
+## QIC: 1498.696
+

The overdispersion parameter is quite small and very close to zero, +the limiting case of a Poisson distribution. Its standard error is not +available in analytic form, but requires bootstrapping.

+

I have stopped this, as even with moderate bootstrap sample sizes +(500), the computation time is prohobitivly long.

+

Diagnostics:

+
acf(residuals(uks_tsc_fit), main = "ACF of response residuals")
+

+
pit(uks_tsc_fit, ylim = c(0, 1.5), main = "PIT Histogram")
+

+
cat("Mean of Pearson residuals:", mean(residuals(uks_tsc_fit,type="pearson")))
+
## Mean of Pearson residuals: -0.0006482209
+
cat("Variance of Pearson residuals: ", var(residuals(uks_tsc_fit,type = "pearson")))
+
## Variance of Pearson residuals:  0.9789472
+
scoring(uks_tsc_fit)
+
## logarithmic   quadratic   spherical    rankprob      dawseb      normsq 
+##  2.55022104 -0.09559065 -0.30694330  1.82121573  3.35794571  0.97552471 
+##     sqerror 
+## 11.54727482
+

One-step ahead prediction:

+
sin_xf <- sin(2*pi*(n+1)/12)
+cos_xf <- cos(2*pi*(n+1)/12)
+
+trendf <- ((n+1):(n+1) - n/2) / n #adjusted to a vector
+
+constf <- 1
+
+xcastk <- cbind(constf, trendf, cos_xf, sin_xf) 
+
+tsglm_poi_f <- predict(uks_tsc_fit,n.ahead = 1, newxreg = xcastk)
+
+tsglm_poi_f$pred
+
## [1] 4.648801
+
tsglm_poi_f$interval
+
##      lower upper
+## [1,]     1    10
+

There is no obvious way for me to improve the models fit.

+

I tried to fit an INGARCH(1,0) as the alpha-parameter is only +borderline significant, but that leads to issues with the +variance-covariance matrix of the parameter vector and worse values for +the scoring rules.

+

I infer from this that the version of the INGARCH model with a +log-link and neg. binomial conditional distriburtion can fit the UK +strikes data comparably with to the GPAR2 model used in \(coconots\), but based on the scroring +rules, the latter would be preferred.

+
+
+

Rig Count Data

+
+

Data span 2014 to 2020.

+

I first use the dataset 2014 to 2020 (March to March), which I +propose to use in the JSS paper.

+
data2 <- read_excel("rigcountsAlaskaL2014-20-march2march.xlsx")
+

Preparation of the regressor. I start with the quarter 1 - dummy +only.

+
Xrc2 <- data2$AlaskaL
+nrc2 <- length(Xrc2)
+
+constrc2 <- rep(1, nrc2)
+
+q1rc2 <- data2$Q1
+
+xregrc2 <- cbind(constrc2, q1rc2)
+

Then I fit a INGARCH(1,1) model with a linear link first:

+
fit_rc2_reg1 <- tsglm(Xrc2, model = list(past_obs = c(1),past_mean=1), xreg = xregrc2)
+
summary(fit_rc2_reg1)
+
## 
+## Call:
+## tsglm(ts = Xrc2, model = list(past_obs = c(1), past_mean = 1), 
+##     xreg = xregrc2)
+## 
+## Coefficients:
+##              Estimate  Std.Error  CI(lower)  CI(upper)
+## (Intercept)  0.629584      0.719     -0.779      2.038
+## beta_1       0.552083      0.108      0.341      0.764
+## alpha_1      0.362488      0.126      0.115      0.610
+## constrc2     0.000452      0.626     -1.226      1.227
+## q1rc2        0.299250      0.271     -0.232      0.830
+## Standard errors and confidence intervals (level =  95 %) obtained
+## by normal approximation.
+## 
+## Link function: identity 
+## Distribution family: poisson 
+## Number of coefficients: 5 
+## Log-likelihood: -641.2468 
+## AIC: 1292.494 
+## BIC: 1311.225 
+## QIC: 1293.475
+

According to the model fit, there is no need for the Q1-dummy.

+

Model diagnostics:

+
acf(residuals(fit_rc2_reg1), main = "ACF of response residuals")
+

+
cat("Mean of Pearson residuals: ", mean(residuals(fit_rc2_reg1,type="pearson")))
+
## Mean of Pearson residuals:  -0.01929286
+
cat("Variance of Pearson residuals: ", var(residuals(fit_rc2_reg1,type = "pearson")))
+
## Variance of Pearson residuals:  0.2370493
+
pit(fit_rc2_reg1, ylim = c(0, 1.5), main = "PIT Histogram")
+

+
scoring(fit_rc2_reg1)
+
## logarithmic   quadratic   spherical    rankprob      dawseb      normsq 
+##   2.0487119  -0.1595609  -0.4074842   0.8768263   2.2563953   0.2366642 
+##     sqerror 
+##   1.7873130
+

The model captures the serial correlation in the data surprisingly +well!

+

One-step ahead prediction:

+
constf <- 1
+q1f <- 0 
+
+xcastf <- cbind(constf, q1f) 
+
+tsglm_rc2_f <- predict(fit_rc2_reg1 ,n.ahead = 1, newxreg = xcastf)
+
+tsglm_rc2_f$pred
+
## [1] 9.271769
+
tsglm_rc2_f$interval
+
##      lower upper
+## [1,]     4    16
+

In an effort to improve the model, I now fit an INGARCH(1,14;,1) +model without any regressors. The lag 14 dependent variable, is trying +to capture potential stochastic seasonality:

+
fit_rc2_reg2 <- tsglm(Xrc2, model = list(past_obs = c(1,14),past_mean=1))
+
summary(fit_rc2_reg2)
+
## 
+## Call:
+## tsglm(ts = Xrc2, model = list(past_obs = c(1, 14), past_mean = 1))
+## 
+## Coefficients:
+##              Estimate  Std.Error  CI(lower)  CI(upper)
+## (Intercept)    0.3878     0.4038    -0.4036      1.179
+## beta_1         0.6139     0.1107     0.3969      0.831
+## beta_14        0.0354     0.0524    -0.0672      0.138
+## alpha_1        0.3081     0.1297     0.0538      0.562
+## Standard errors and confidence intervals (level =  95 %) obtained
+## by normal approximation.
+## 
+## Link function: identity 
+## Distribution family: poisson 
+## Number of coefficients: 4 
+## Log-likelihood: -641.2983 
+## AIC: 1290.597 
+## BIC: 1305.581 
+## QIC: 1290.692
+

The parameter related to the stochastic seasonality is not +statistically significant.

+

Model diagnostics:

+
acf(residuals(fit_rc2_reg2), main = "ACF of response residuals")
+

+
cat("Mean of Pearson residuals: ", mean(residuals(fit_rc2_reg2,type="pearson")))
+
## Mean of Pearson residuals:  -0.02644024
+
cat("Variance of Pearson residuals: ", var(residuals(fit_rc2_reg2,type = "pearson")))
+
## Variance of Pearson residuals:  0.2384399
+
pit(fit_rc2_reg2, ylim = c(0, 1.5), main = "PIT Histogram")
+

+
scoring(fit_rc2_reg2)
+
## logarithmic   quadratic   spherical    rankprob      dawseb      normsq 
+##   2.0488764  -0.1595560  -0.4076225   0.8765324   2.2593817   0.2383772 
+##     sqerror 
+##   1.7810585
+

The residuals seem to look a bit ‘better’ in some sense. But I fail +to see any real improvement here.

+

I see no obvious way to improve the model.

+
+
+
+

+
+

Time span 2017 to 2024

+

I also include the analysis based on the original time span for the +rig counts, including the Covid19 period here.

+
datarc <- read_excel("rigcountsAlaskaL2017-24.xlsx")
+
+mean(datarc$AlaskaL)
+
## [1] 6.394521
+
var(datarc$AlaskaL)
+
## [1] 5.794475
+
plot(datarc$AlaskaL, type="l")
+

+
forecast::Acf(datarc$AlaskaL)
+

+
forecast::Pacf(datarc$AlaskaL)
+

+

I fit a model with a Q1-dummy and a linear trend as we use it in the +coconots package (note log-link is required here):

+
Xrc <- datarc$AlaskaL
+nrc <- length(Xrc)
+
+trendrc <- (1:nrc - nrc/2) / nrc
+constrc <- rep(1, nrc)
+
+q1rc <- datarc$Q1
+
+xregrc <- cbind(constrc, trendrc, q1rc)
+
+
+fit_rc_reg1 <- tsglm(Xrc, model = list(past_obs = c(1),past_mean=1), link="log", distr = "poisson", xreg = xregrc)
+
+
+summary(fit_rc_reg1)
+
## 
+## Call:
+## tsglm(ts = Xrc, model = list(past_obs = c(1), past_mean = 1), 
+##     xreg = xregrc, link = "log", distr = "poisson")
+## 
+## Coefficients:
+##              Estimate  Std.Error  CI(lower)  CI(upper)
+## (Intercept)    1.2061     0.9444   -0.64496      3.057
+## beta_1         0.7641     0.1285    0.51234      1.016
+## alpha_1       -0.0900     0.1689   -0.42098      0.241
+## constrc       -0.7297     1.0240   -2.73669      1.277
+## trendrc        0.1570     0.0838   -0.00734      0.321
+## q1rc           0.0745     0.0532   -0.02981      0.179
+## Standard errors and confidence intervals (level =  95 %) obtained
+## by normal approximation.
+## 
+## Link function: log 
+## Distribution family: poisson 
+## Number of coefficients: 6 
+## Log-likelihood: -713.1914 
+## AIC: 1438.383 
+## BIC: 1461.782 
+## QIC: 1438.979
+

Hm, the alpha-parameter related to the lagged conditional mean seems +not to be significant; the trend term wants to be positive (!) and the +parameter associated with the first quarter is not significant.

+
acf(residuals(fit_rc_reg1), main = "ACF of response residuals")
+

+
pit(fit_rc_reg1, ylim = c(0, 1.5), main = "PIT Histogram")
+

+
mean(residuals(fit_rc_reg1,type="pearson"))
+
## [1] -0.02232376
+
var(residuals(fit_rc_reg1,type = "pearson"))
+
## [1] 0.2603702
+
scoring(fit_rc_reg1)
+
## logarithmic   quadratic   spherical    rankprob      dawseb      normsq 
+##   1.9539492  -0.1781824  -0.4303744   0.8066894   2.0772362   0.2601552 
+##     sqerror 
+##   1.7539446
+
+
+
+

+

Finally, I try to include an intervention term, a dummy taking on the +value 1 from the second quarter or 2020 onwards to check, if the Covid19 +period want to be treated different from the previous time period.

+
intervention <- interv_covariate(nrc , tau = c(159), delta = c(1))
+fit_rc_regi <- tsglm(Xrc, model = list(past_obs = c(1),past_mean=1), link= "log", distr = "poisson", xreg = intervention)
+
+summary(fit_rc_regi)
+
## 
+## Call:
+## tsglm(ts = Xrc, model = list(past_obs = c(1), past_mean = 1), 
+##     xreg = intervention, link = "log", distr = "poisson")
+## 
+## Coefficients:
+##              Estimate  Std.Error  CI(lower)  CI(upper)
+## (Intercept)    0.0920     0.1235    -0.1501     0.3341
+## beta_1         0.7842     0.1368     0.5160     1.0524
+## alpha_1        0.1222     0.1523    -0.1762     0.4206
+## interv_1      -0.0268     0.0364    -0.0982     0.0446
+## Standard errors and confidence intervals (level =  95 %) obtained
+## by normal approximation.
+## 
+## Link function: log 
+## Distribution family: poisson 
+## Number of coefficients: 4 
+## Log-likelihood: -703.6384 
+## AIC: 1415.277 
+## BIC: 1430.876 
+## QIC: 1414.709
+

The intervention dummy is nowhere near a sensible significance +level.

+
acf(residuals(fit_rc_regi), main = "ACF of response residuals")
+

+
pit(fit_rc_regi, ylim = c(0, 1.5), main = "PIT Histogram")
+

+
mean(residuals(fit_rc_regi,type="pearson"))
+
## [1] -0.04130746
+
var(residuals(fit_rc_regi,type = "pearson"))
+
## [1] 0.2137347
+
scoring(fit_rc_regi)
+
## logarithmic   quadratic   spherical    rankprob      dawseb      normsq 
+##   1.9277765  -0.1843596  -0.4387367   0.7609530   2.0325084   0.2148555 
+##     sqerror 
+##   1.3042387
+
+ + + + +
+ + + + + + + + + + + + + + + diff --git a/Rnotebook-tscounts.md b/Rnotebook-tscounts.md new file mode 100644 index 0000000..d2d057b --- /dev/null +++ b/Rnotebook-tscounts.md @@ -0,0 +1,460 @@ +THIS VERSION: 6.12.2024 + + + +In this document I try to fit the UK strikes data and the rig count data +that we propose in the JSS paper to be analysed with the +*c**o**c**o**n**o**t**s* package with the *t**s**c**o**u**n**t* package. + +I start with the UK strikes data. + +Then I present an analysis of the rig counts in the time span 2014 to +2020. The tscount package does a remarkably good job in capturing the +dependence structure, but struggles with the slight underdispersion in +the data. Values of the scoring rules are considerably larger (=worse) +as those from the GPAR2 model. + +Finally, I then present results for rig count in the time span 2017 to +2024. Here I find no specification of the models in tscount, that can +capture the dynamics in the data satisfactorily. + +# UK Strikes Data + + data_uks <- read_excel("data_strikes.xlsx") + +I start with the UK strikes data (public sector). In +*c**o**c**o**n**o**t**s*, we propose a GPAR2 model with harmonics and a +linear trend (using a log-link). + +I generate a holdout sample of 3, and prepare the regressors first: + + hos <- 3 + + X <- data_uks$public + n <- length(X) - hos + + sin_x <- sin(2*pi*1:n/12) + cos_x <- cos(2*pi*1:n/12) + trend <- (1:n - n/2) / n + const <- rep(1, n) + xregtsc <- cbind(const, trend, cos_x, sin_x) + +Then we fit an INGARCH(1,1) type of model with a log link and negative +binomial conditional distributional (Poisson led to a slightly u-shaped +PIT histogram) assumption: + + uks_tsc_fit <- tsglm(data_uks$public[1:n], model = list(past_obs = c(1),past_mean=1), link = "log", distr = "nbinom", xreg = xregtsc) + +The standard regression output: + + summary(uks_tsc_fit) + + ## + ## Call: + ## tsglm(ts = data_uks$public[1:n], model = list(past_obs = c(1), + ## past_mean = 1), xreg = xregtsc, link = "log", distr = "nbinom") + ## + ## Coefficients: + ## Estimate Std.Error CI(lower) CI(upper) + ## (Intercept) 1.2547 0.8288 -0.3697 2.879134 + ## beta_1 0.3990 0.0670 0.2677 0.530197 + ## alpha_1 0.2153 0.1382 -0.0556 0.486154 + ## const -0.4992 0.6884 -1.8485 0.850020 + ## trend -0.3439 0.1196 -0.5783 -0.109619 + ## cos_x -0.0609 0.0309 -0.1214 -0.000343 + ## sin_x 0.0165 0.0306 -0.0434 0.076477 + ## sigmasq 0.0547 NA NA NA + ## Standard errors and confidence intervals (level = 95 %) obtained + ## by normal approximation. + ## + ## Link function: log + ## Distribution family: nbinom (with overdispersion coefficient 'sigmasq') + ## Number of coefficients: 8 + ## Log-likelihood: -729.3632 + ## AIC: 1474.726 + ## BIC: 1503.974 + ## QIC: 1498.696 + +The overdispersion parameter is quite small and very close to zero, the +limiting case of a Poisson distribution. Its standard error is not +available in analytic form, but requires bootstrapping. + +I have stopped this, as even with moderate bootstrap sample sizes (500), +the computation time is prohobitivly long. + +Diagnostics: + + acf(residuals(uks_tsc_fit), main = "ACF of response residuals") + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-6-1.png) + + pit(uks_tsc_fit, ylim = c(0, 1.5), main = "PIT Histogram") + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-6-2.png) + + cat("Mean of Pearson residuals:", mean(residuals(uks_tsc_fit,type="pearson"))) + + ## Mean of Pearson residuals: -0.0006482209 + + cat("Variance of Pearson residuals: ", var(residuals(uks_tsc_fit,type = "pearson"))) + + ## Variance of Pearson residuals: 0.9789472 + + scoring(uks_tsc_fit) + + ## logarithmic quadratic spherical rankprob dawseb normsq + ## 2.55022104 -0.09559065 -0.30694330 1.82121573 3.35794571 0.97552471 + ## sqerror + ## 11.54727482 + +One-step ahead prediction: + + sin_xf <- sin(2*pi*(n+1)/12) + cos_xf <- cos(2*pi*(n+1)/12) + + trendf <- ((n+1):(n+1) - n/2) / n #adjusted to a vector + + constf <- 1 + + xcastk <- cbind(constf, trendf, cos_xf, sin_xf) + + tsglm_poi_f <- predict(uks_tsc_fit,n.ahead = 1, newxreg = xcastk) + + tsglm_poi_f$pred + + ## [1] 4.648801 + + tsglm_poi_f$interval + + ## lower upper + ## [1,] 1 10 + +There is no obvious way for me to improve the models fit. + +I tried to fit an INGARCH(1,0) as the alpha-parameter is only borderline +significant, but that leads to issues with the variance-covariance +matrix of the parameter vector and worse values for the scoring rules. + +I infer from this that the version of the INGARCH model with a log-link +and neg. binomial conditional distriburtion can fit the UK strikes data +comparably with to the GPAR2 model used in *c**o**c**o**n**o**t**s*, but +based on the scroring rules, the latter would be preferred. + +# Rig Count Data + +## Data span 2014 to 2020. + +I first use the dataset 2014 to 2020 (March to March), which I propose +to use in the JSS paper. + + data2 <- read_excel("rigcountsAlaskaL2014-20-march2march.xlsx") + +Preparation of the regressor. I start with the quarter 1 - dummy only. + + Xrc2 <- data2$AlaskaL + nrc2 <- length(Xrc2) + + constrc2 <- rep(1, nrc2) + + q1rc2 <- data2$Q1 + + xregrc2 <- cbind(constrc2, q1rc2) + +Then I fit a INGARCH(1,1) model with a linear link first: + + fit_rc2_reg1 <- tsglm(Xrc2, model = list(past_obs = c(1),past_mean=1), xreg = xregrc2) + + summary(fit_rc2_reg1) + + ## + ## Call: + ## tsglm(ts = Xrc2, model = list(past_obs = c(1), past_mean = 1), + ## xreg = xregrc2) + ## + ## Coefficients: + ## Estimate Std.Error CI(lower) CI(upper) + ## (Intercept) 0.629584 0.719 -0.779 2.038 + ## beta_1 0.552083 0.108 0.341 0.764 + ## alpha_1 0.362488 0.126 0.115 0.610 + ## constrc2 0.000452 0.626 -1.226 1.227 + ## q1rc2 0.299250 0.271 -0.232 0.830 + ## Standard errors and confidence intervals (level = 95 %) obtained + ## by normal approximation. + ## + ## Link function: identity + ## Distribution family: poisson + ## Number of coefficients: 5 + ## Log-likelihood: -641.2468 + ## AIC: 1292.494 + ## BIC: 1311.225 + ## QIC: 1293.475 + +According to the model fit, there is no need for the Q1-dummy. + +Model diagnostics: + + acf(residuals(fit_rc2_reg1), main = "ACF of response residuals") + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-12-1.png) + + cat("Mean of Pearson residuals: ", mean(residuals(fit_rc2_reg1,type="pearson"))) + + ## Mean of Pearson residuals: -0.01929286 + + cat("Variance of Pearson residuals: ", var(residuals(fit_rc2_reg1,type = "pearson"))) + + ## Variance of Pearson residuals: 0.2370493 + + pit(fit_rc2_reg1, ylim = c(0, 1.5), main = "PIT Histogram") + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-12-2.png) + + scoring(fit_rc2_reg1) + + ## logarithmic quadratic spherical rankprob dawseb normsq + ## 2.0487119 -0.1595609 -0.4074842 0.8768263 2.2563953 0.2366642 + ## sqerror + ## 1.7873130 + +The model captures the serial correlation in the data surprisingly well! + +One-step ahead prediction: + + constf <- 1 + q1f <- 0 + + xcastf <- cbind(constf, q1f) + + tsglm_rc2_f <- predict(fit_rc2_reg1 ,n.ahead = 1, newxreg = xcastf) + + tsglm_rc2_f$pred + + ## [1] 9.271769 + + tsglm_rc2_f$interval + + ## lower upper + ## [1,] 4 16 + +In an effort to improve the model, I now fit an INGARCH(1,14;,1) model +without any regressors. The lag 14 dependent variable, is trying to +capture potential stochastic seasonality: + + fit_rc2_reg2 <- tsglm(Xrc2, model = list(past_obs = c(1,14),past_mean=1)) + + summary(fit_rc2_reg2) + + ## + ## Call: + ## tsglm(ts = Xrc2, model = list(past_obs = c(1, 14), past_mean = 1)) + ## + ## Coefficients: + ## Estimate Std.Error CI(lower) CI(upper) + ## (Intercept) 0.3878 0.4038 -0.4036 1.179 + ## beta_1 0.6139 0.1107 0.3969 0.831 + ## beta_14 0.0354 0.0524 -0.0672 0.138 + ## alpha_1 0.3081 0.1297 0.0538 0.562 + ## Standard errors and confidence intervals (level = 95 %) obtained + ## by normal approximation. + ## + ## Link function: identity + ## Distribution family: poisson + ## Number of coefficients: 4 + ## Log-likelihood: -641.2983 + ## AIC: 1290.597 + ## BIC: 1305.581 + ## QIC: 1290.692 + +The parameter related to the stochastic seasonality is not statistically +significant. + +Model diagnostics: + + acf(residuals(fit_rc2_reg2), main = "ACF of response residuals") + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-16-1.png) + + cat("Mean of Pearson residuals: ", mean(residuals(fit_rc2_reg2,type="pearson"))) + + ## Mean of Pearson residuals: -0.02644024 + + cat("Variance of Pearson residuals: ", var(residuals(fit_rc2_reg2,type = "pearson"))) + + ## Variance of Pearson residuals: 0.2384399 + + pit(fit_rc2_reg2, ylim = c(0, 1.5), main = "PIT Histogram") + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-16-2.png) + + scoring(fit_rc2_reg2) + + ## logarithmic quadratic spherical rankprob dawseb normsq + ## 2.0488764 -0.1595560 -0.4076225 0.8765324 2.2593817 0.2383772 + ## sqerror + ## 1.7810585 + +The residuals seem to look a bit ‘better’ in some sense. But I fail to +see any real improvement here. + +I see no obvious way to improve the model. + +# + +## Time span 2017 to 2024 + +I also include the analysis based on the original time span for the rig +counts, including the Covid19 period here. + + datarc <- read_excel("rigcountsAlaskaL2017-24.xlsx") + + mean(datarc$AlaskaL) + + ## [1] 6.394521 + + var(datarc$AlaskaL) + + ## [1] 5.794475 + + plot(datarc$AlaskaL, type="l") + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-1.png) + + forecast::Acf(datarc$AlaskaL) + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-2.png) + + forecast::Pacf(datarc$AlaskaL) + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-3.png) + +I fit a model with a Q1-dummy and a linear trend as we use it in the +coconots package (note log-link is required here): + + Xrc <- datarc$AlaskaL + nrc <- length(Xrc) + + trendrc <- (1:nrc - nrc/2) / nrc + constrc <- rep(1, nrc) + + q1rc <- datarc$Q1 + + xregrc <- cbind(constrc, trendrc, q1rc) + + + fit_rc_reg1 <- tsglm(Xrc, model = list(past_obs = c(1),past_mean=1), link="log", distr = "poisson", xreg = xregrc) + + + summary(fit_rc_reg1) + + ## + ## Call: + ## tsglm(ts = Xrc, model = list(past_obs = c(1), past_mean = 1), + ## xreg = xregrc, link = "log", distr = "poisson") + ## + ## Coefficients: + ## Estimate Std.Error CI(lower) CI(upper) + ## (Intercept) 1.2061 0.9444 -0.64496 3.057 + ## beta_1 0.7641 0.1285 0.51234 1.016 + ## alpha_1 -0.0900 0.1689 -0.42098 0.241 + ## constrc -0.7297 1.0240 -2.73669 1.277 + ## trendrc 0.1570 0.0838 -0.00734 0.321 + ## q1rc 0.0745 0.0532 -0.02981 0.179 + ## Standard errors and confidence intervals (level = 95 %) obtained + ## by normal approximation. + ## + ## Link function: log + ## Distribution family: poisson + ## Number of coefficients: 6 + ## Log-likelihood: -713.1914 + ## AIC: 1438.383 + ## BIC: 1461.782 + ## QIC: 1438.979 + +Hm, the alpha-parameter related to the lagged conditional mean seems not +to be significant; the trend term wants to be positive (!) and the +parameter associated with the first quarter is not significant. + + acf(residuals(fit_rc_reg1), main = "ACF of response residuals") + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-19-1.png) + + pit(fit_rc_reg1, ylim = c(0, 1.5), main = "PIT Histogram") + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-19-2.png) + + mean(residuals(fit_rc_reg1,type="pearson")) + + ## [1] -0.02232376 + + var(residuals(fit_rc_reg1,type = "pearson")) + + ## [1] 0.2603702 + + scoring(fit_rc_reg1) + + ## logarithmic quadratic spherical rankprob dawseb normsq + ## 1.9539492 -0.1781824 -0.4303744 0.8066894 2.0772362 0.2601552 + ## sqerror + ## 1.7539446 + +# + +Finally, I try to include an intervention term, a dummy taking on the +value 1 from the second quarter or 2020 onwards to check, if the Covid19 +period want to be treated different from the previous time period. + + intervention <- interv_covariate(nrc , tau = c(159), delta = c(1)) + fit_rc_regi <- tsglm(Xrc, model = list(past_obs = c(1),past_mean=1), link= "log", distr = "poisson", xreg = intervention) + + summary(fit_rc_regi) + + ## + ## Call: + ## tsglm(ts = Xrc, model = list(past_obs = c(1), past_mean = 1), + ## xreg = intervention, link = "log", distr = "poisson") + ## + ## Coefficients: + ## Estimate Std.Error CI(lower) CI(upper) + ## (Intercept) 0.0920 0.1235 -0.1501 0.3341 + ## beta_1 0.7842 0.1368 0.5160 1.0524 + ## alpha_1 0.1222 0.1523 -0.1762 0.4206 + ## interv_1 -0.0268 0.0364 -0.0982 0.0446 + ## Standard errors and confidence intervals (level = 95 %) obtained + ## by normal approximation. + ## + ## Link function: log + ## Distribution family: poisson + ## Number of coefficients: 4 + ## Log-likelihood: -703.6384 + ## AIC: 1415.277 + ## BIC: 1430.876 + ## QIC: 1414.709 + +The intervention dummy is nowhere near a sensible significance level. + + acf(residuals(fit_rc_regi), main = "ACF of response residuals") + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-21-1.png) + + pit(fit_rc_regi, ylim = c(0, 1.5), main = "PIT Histogram") + +![](Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-21-2.png) + + mean(residuals(fit_rc_regi,type="pearson")) + + ## [1] -0.04130746 + + var(residuals(fit_rc_regi,type = "pearson")) + + ## [1] 0.2137347 + + scoring(fit_rc_regi) + + ## logarithmic quadratic spherical rankprob dawseb normsq + ## 1.9277765 -0.1843596 -0.4387367 0.7609530 2.0325084 0.2148555 + ## sqerror + ## 1.3042387 diff --git a/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-12-1.png b/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-12-1.png new file mode 100644 index 0000000000000000000000000000000000000000..5e8dea5783eb0c5ab6968b6fd1af9bd62f468b5c GIT binary patch literal 4401 zcmcIo3piAJyPwA7wl{LG#8kqNNQlO5vE7A1(lH!wT z#LQ3$m1K6g#DokZxy+a*mkj1x)AxPP`JQvmKIc5o*?Xsgmj zK5h!~o8=)8h=PZ^vmXQkr9dF-KFUgg5^q0KY4GwAwf~?Cc!fZa5C|Fq5krtj2oeoJ zk|1ahG*AkTkr1>*EIH6fGzkgHNg&vu189o=u0~5VG&Crrfe@3B5E2pznv&2E5*kS& zp-Chfc#>#r5~yXfNo+9<4WXftG&Gt^)worO#`)JHcc#MlaOo@noR@o z-3io44l!FS2K_BkfP+hsi*Y|34}mDQN!XzXJ+tKx4m1lgs=iQDPr?6vLH;F{PWJJ0p1x?XmNPS@o1gi0Qs z{uY^da^c)=^954~<5yrC@lQhJ0m4N4vEjfGezUk|qi{gP5xqWGTt=H|X6~Jcd^jRF zNk1x_C|`q(4h0PgOxO+mPfuPW{2_s z^Z3|;g+<$6fp8-$&6azmN6F*8R-#oIb`{n%GHA0i;?8vdY zIC1%Voitqkhg!?do?V74Vv4YA`itD7?us807ugNs)Cj%95Bz8IS1nLUVU0$#mqS)b z#;)vx)uUhDTv%5;A0L}K&3oXOtHgVT?Gu#%LNqj~rO!COfdb?h)E{YmA07Ch%!0gh zN+wa99AwvRp7AR3`neF3rzu{WD&hcFc76Z}6?VGwaXUsZfTD$Vag2mD4#aY%nJdz4p zt^el6^*yTf@yt06F15S2IdQRXt4;j`27_?F)M_qC={qbWI@&nWEh2SaK zJ1Mq1fHIF8I<`96k*JLk6%;eLu)+KKk>(FFZgUPbXiW+%C+V#v4r!}ZW)~FZsGOplz`Z?*)psK-Lr z)*or?jn**q?fA;P`jz}fi5%^ac2XEq+p}djKGf=oc==xQw*g3RC4FAwdf!$znc5JK z<>^P?mO4)@zFWJ5Uki#?OP}}dn~|r67d8C`D}J)mG;i!Q47m|XBmgo$QOq_>{vvB3 z4b8Ini$NtwS)MDfCiDf4e_p+h4R|r9V!*oJ>>4wiWEcm{DnJl;QDo!#(&`yy1jAV_ zfgV-UN?trYE~gfJD8uQDdSJQ_W$u}y%57!DUkDfZgs|7AlCb;v9q&~TjOTLH$DS+g_p>_5eoM6CYdLcAcexrSCHv(R^RzADnb&RA`AVcinU+tFt}~+m zsU1A!OiGT{h)u4SJ^wf9Eb1Wn{6T7+UZGQ&l8CWKc{tu_vpIc21a@{{XQDO zWmO_nqx(xYVNsbms*^Z;N=S{9u(@If$c{sIisF*4U+MK|RYo*fE6GQ|KYZgtDNMM#%?k#|#&~EX^Z42KDROE-V0ixy z1DP{RkhbCOAY^JVm}G+~$@b5VgUo2VZlhBxmGl!>8$JpNbR1m6FI_tUH0rM+L`|7= zPRApXBkr zc7$i0FAb(Tb(6(O*hVXPp$<@)CssKaEz)6knu(p|pCGD0Dck(9W?sOGFMo-f{&35;f4X(s%24Up@nk@dZEcV|Da_ z`;OO*Kjnq0I{b&7^4+E^2S3<7UZ&fr+$(UxJ+F;-Ib*(*DcOXNHqt-dS%MIA0h#8o zCo$PdtjYgH7ogMd(>b@H+ zw*GTa7EUo8+N{TNbbub)L(0rh!Mq@uYJJ@<51U%$%k`*YS?gr*T8Y{(-V4jl0U>e2WS13U36yYV#AwrG zQM87@3I_&)uMVYrYVP?QJ)`c=$kbR)`iJIkY(>%_bW7)0Fo!Ez5&n`Nwd>I4i)Ic(DNLN)=YKXx}ydGZ`+ zs;kRd*A-Xp>PnFUJ&2b98H>)W&6@~{n16E&TfcnZc$9=NX4SRLE_Kpv8{a?w=?|zU z;o~5N01bAPPoVDe>$d2wf?1t|y*e27dYlLpow{7aP)7I@&GN0FC-0e|zqnHNTej~X zmc3ThUiearu%~yUd5SKoa&wq|^!XN*7~ucVur=&h9}08m@AK7Y(;RVEljrVtW4`xFcs}$>Wv`j^1p1&!g@{@4-6F$Ruom?#?yn zTf3m42aPMi*jWqA$d}f8g@L3EaR@$ zVlw$*0n97F_d4eHnPT(Ncif#g3b)P`(MqRfC<`Y2!_G##Y2`@oM8Ilz~;#yQ??|LMFQq{02 z%E0L1IC;tMgG*;;&_!+}$Df~Qu&@8x)Yb^G1*#zaVNf; z4L89j&~Y#CnTZwep$YdXo%#6--?Kl4EDiPcm@+DS2T&rei74#EtJ4GB@&x^>QRj0Y5AHq8Of^;4;LTj8mC{*{~i4~0Hpu` literal 0 HcmV?d00001 diff --git a/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-12-2.png b/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-12-2.png new file mode 100644 index 0000000000000000000000000000000000000000..3bc2d5a2bfbddb3f665c1f8b3dc425798975a855 GIT binary patch literal 6120 zcmeHL2UL^UwobsHpp@r;jS?9|u>gZ8O-K<0krLDqq^n2(F+pmmNd!h^1c_w?r6h_P z6p;Z!=%6ydC_;h(kpzsi5P?vH&_nK@IL?~;)}6cFz3;8}?py2rIZ4jhXYYM>`S$;x z|D?D$+bJq+Q2>EJiuQ+Xj)6cj6c7mfag8iMaXMxw2mCd-IDUT!_y>V3Kp->-BnDYn zfGp4;3lazoG&(>9#ugy7geGqjt0@u7JvYq^hE-gUKeKmx+-F^uVp5t0#T^^rLhQUx25?grAn0ULY@?=)ALe)j%BT>^J$T z3g7>*2>Pz`y#jLGeZ@7`W!mME4_^$FJ@^1Y18X1K$|kDpRKXyws9rJIYNKGIuJSG9 z|A=j}^dQ4Gp&Lt{2o+U$pXq-T{1)CLs=KrDerg#Chs@WlQ^63)MwE(Jy|w(V3Ai5R z;Z`;XejJh2Q@fpFc=8}vU#?wl-d(t`SQQv`uex>sxuL#nwMf*ACNFjQ(i5^hQ^*9x z4nvoTp5q#8I^$LXWWYypC%P3eU&Mj6O@Jl9RcAvM54dfP?Q6}3>N zNjq0GZMYUPqXMiSP=?F75vo=%S{!IzxywHh$AqP*LuS?hQ`CJF0T({jW`-gsMS;C> z)cXU6Zpc_92Ffn0LTIlz+^FV-@z=F!A3qh$hbqU4=7m1P6gx@u{NtEh=Z&ryjhhom z&cl?d!?E)B6(^5JUlCi!$+7V-9*E!bTPE$_!e=0@*7b{@`**EW6EFvkBjymai*s*2 z&j+LWAy-T$4T9QW;zyhV_<;i5evplNbjDL$i0JA*U<$UdILhfOBwPnV@-G1H=jyWc zEmX>*2isx;&m(L(t@phhB1O+8@Q;WVNvU*igqz@QL004NbgSWdNXO>0G>oi;O}Nk} zL(UbEgS6$CP*COGh^`6Y)*`6@J82hXd}|(tkj2juL|4ve9R=8Yrnl$0DYyQx9M?*f z8Bx5iSYxE>5}<ayNypUEk$pHkchuK-fuG7YFgzA@$YfHgQ$ zt|P38LR=oyqwFVov~40wI>0Jo`(uohf_aU~XjK9S8oEF_t z&mMktD3AC#sfHK)p#E|8&W))uua)zMQja1Qz-8G+3VLwLZDU*T&}v`-KuV$rb`*lDL)}rC1p45URUjsleDEvkD~MIxmx!bx2YYo zWCgz0eg*!Tze1JXZqHAnkrv@ZPz&u!45343zY{;?o$f zhK+|3|NFq^mB@G?W>(e6096p~4%XL8$tAjhzSP2Z*z+6Arkx({sWiO^1STr0fcIeV zmmf`Z!$LrSff~yva*k$WUL*r^*H(NUzC*QT3`nYLiSRTl0{;QVC3~swL0ywLk6t zrYQ@d$vJwqWVQA$hL$q5@w`pIjp-Ddz2}1PLR8v>TT3nEk9UxgDud^`ateWpne^NV z?yOx7O{$Pq_VfsEtvN36i!XVitwFZtgs-CnXv7~zvx<*&zRm@W_Bajh9qPH4c2UMd zwkge2{DVJRVgb>^pdgrR-fDOh5ezExa_xeJ-}Su@#Hkk@PO8Uyy{qj`F4ccSk;*+P z6m~W|5t*<&y(=!7NUYmrnmaK;cS3N-=a50a*_4o1r^wF|*;--NsWP{GhfR%R`ghjf zZRp;BW;RJ8%W@?sRXvHT)(Y^yGiFjEBt3HzjznfKWBL&9aHqE8Q zKT=>!U681z+-@0R2RKhV01-gwXLOb^vUwSQC)}WOl~11~@b>+e7x*=hLnw^NS*(HC zVbNdmFL?%YhC-dWq1u0-VGx%CpE)lt4Qf>0`I7y;Y&uwTrE~akNE+|g)^*Y>k=kA3 zTp%5qXK8}^ck!i8J+}>h*HOdkbsEZAyKsCo{p{;jc`5&&`Gj!qek`<87~&D7@W{0D zKeM57ukvr8`cp4UJm6fidyh5>>-xo)>?YP?%MR|E)*D$HrFjJ&;c-Pp;DiqQt~8m& zyShzbgXl}rHGT~&u*#PiQ3~6FnzQdP*Yh=1;764^4!q**9D3EN@ORDqBjW$mxIe=0 zAIh<|QXqJob>wUMRaVm|za)4(mZF)myuP2NcB=xlTUvDA-10q`O;21u7xYkyoW#!# zSAcy^Kdew%2#zjz`B1jPksj(#(t4GDZK&<+iA-6d9g~6%xuE+?&~Cj@va56D4IDpy zyIy^=83v+j>@+!g^ZCx^0{}x!n<_I*liVJ)voG=y@%HF^D&0Z8$QVNNY7U@zABoP{ zNg1+$t0Nk+RWH5ki?vLBbnOmADCFCv_#(&?)RWJiL46H> zrCeig zKDk)VMOdWljmp^WD2MOeSIwwfqeW+OCqn4kzbO?&-OXG2aiKZ&nv&V!tL_nX+UI}H zgN^$~ijoc?lrN7loZ7Bji#1xu--2W9Dp8wGH@6b7lqX{eG)321(WS|^H5okA&2VNC z_8oKucRK7~!u**%%2TU#==#}&H$j19yq(Edo6^d6qYnwcoI0i)7nJU{9c;x%cWObG zG_y||UUDnq)Y`bdf%Ep!P%elr*RI3WKJ3ninliEe5lr)VBf%8!)4M?PGb6+%jeg8+ zJbuArOnFKq?@9QK?z-CDh+Hkyh7Eoj?tTAojQ8*(!@f8yE?g%4MbMEFM*S9G;pm@MNDe=IxS3LCNHqcOrt0n*T+z=v*) z*gXuL_E>+$=br<`|KMrXnt<*i{>+)7vJ8iC=XBa`#(r_+T4tfLhcNi9;!{RZnJFF4 zjqoo|Ucdi}iHdW*(tC7hjmF7uDI9@nVUms3jp$w0{^#Zm1|Acg*LK#P)FA0oTydOl zSGUW(+$_U>5HkI#dA1GGP;wbnPQV?|aLo`UzGmrETgS(lR*XWcgplg+)Rh-D~St(aGVltO?OvtFTRg z&eKK*-gp)h9X2)gpaRqp|rPV&m-V78{{apFYK_tT067W-Au(D%Qp&)e}@3l+n z;{J(RBAuGi3o90NXWN%@<1nQwy&0TPCvHIwno(GxaTUjzDop#&W|YgQ)Xt>$AI$S*7V%~Z{Js;*v1Y=a zJ$G+*e7c@QjJl@@A2MIZ@4aBxfL5-+slf$0q2+7&y|e15eef#p$;?;1Cu$_3O8mYh zN<&OvXG2LZRVjUp&{27wJC==IQx0XWJf2Npj!%B@N==wATG1rt3`VFl3!%O+qufU4 z3A@7dBA%+t)RtVlu!^(Yz4xsk5lOpv^;G0+(r2vU92V|U1ZtNvAW}JQJvb~TT$X(m gz~KMJ(V1A;QE)NUIEaxdIkC4teH9ae8OzhUBJiylcj7D-z9SH51cHD-2oPvA0!=`m$p``r z3e3VX8bMg4S1$sZKt{tn83qG(fK3T&YQn09LV!sM3;`L9AfwT+DVcyE6VPNbflQ{r zA2NkOhP4a^nIWJM5EKHMLLgAcXbPD?p}?-No+qMR zKnc*UfPRVpIKb2Mo*{FRq~e)BFKqJRs}xm#w7n$fi*40%X%;8JK5asZR8E`)lPt%#EnJ+SB*x$Mc3{IO1qY(rV4x&Tn(_Lc+1I?>{zcA8~*8) zkG&$#GchD;mQQJsrCkO*79Kj6)NZ-##T#Pw6T-GG`pt?_YCpNFDhO+E1b^%aChtT+WEw87oVWft~}$6 zGC9|3yrE&(g13!`ZP*_pbUpISBaxGoCJcm3UW$ZJa~6iWiE%x|8}|6{Pdm&qn+xx! zdNeZJ&*9h8J*5nm01qkm8G?)}yKQW6*e&l<2yL0*#P=BE2dAE3n7W$Kd}ibi1YR4} zYH2d2eiEimCcFcvPQG`U|;R)vWy1B3e%S1_8qiGzo2{!A6KE7Y&$cl zARXG^S?77Py1uDnDPx!z$q*+Iul5TWFANet9?<<1TdDdOKfc7bzgQhr61P9lSG=Nk zLg&2KNF;XVqn_uK20w4Qetfj4O~aIaD-JRL-5@dXfG*SG z$3#{j*SU1bP4L!KP>hbIF77BK36DQK(RT?M_cgidowor}X#;p0fWlR2o5cL~p%jte zA;teOdS8Kxua2hyt=cmcq1^|yF9(RrxSqCCH}b5N{TY07oD5VI?QVTEvO5?hp^HKG z=@dTU++DOiquC&Q~=;;e%g~yt&K8TFHq=H#Aq(JaKPN;hZygCe*MKPmgbj8 z>~>R~n|2q%57`!k&>Xk4m}>nH_midC)I3~BncE4zk@;f^c$WV<_TZq2iKaeomfqr) zgqJQyAy)|**Z#45DN%b7j-K&aKnz=hw|!HqZMMGgFckpnbgB8-Fm)kF#rILVwT_+%mz`IVxq}b4#sFt}wy``DEr|MJ6x^^ zH;?H&2|ZkB|De=R^~6kEHrq%_Vs)Iq2qhs;e&aBvQ4^=11-BTi>;hBj-ZqeaH28=_6trV+U1QA+aIdV9g#LSJavVFZ(_BN#6HKS|7#G3@TQK~CjX659U~UO#N~7AtuhLGe zuA(mJC)}(Y=c!d~ye4k;S}4R|W*J3?)de4hSl_+?GQK~CNFu~+BWT(&lQ)oL+2E9Bz)~ydGu}t6* zKJEc2t9a~yt6|3O7Z2P9&9+GFy%`g$wG5s-kepLU?7RApUDO~;kMW3PbXSeokT^$^=4}~;uNxE4zV=9D$;Fr#bjTi zdPVn?ebHg;X(bZC^pn*Q*TXt$C7)#AB3ePpJ46Zc+9K=Xw zkgF!a&04<*qfgn}MP#&kKg@YXyw-d6;@)?b@JO@uYUf+rcFey1n?>y;{KFz|vBa%9 zC(LEJgm~%+Mo+|G`afD{qJEA`G|<()d|wTB?Lb(b*xA0eoL$IshTQ+7m9h(uUtbZM zO)6E91vmcUkkDgc`r65?;0_r7vY=@bYnPLm7E&qTjdJ;!>o<>t;ZW%Bx2o6I+Y-sfu}e@bBqBh7jx!`VBj{i-Q+&7^kMCG!XM$1uugnbY~o1%hJ-@ZqEYP+b4U zU7_TfJBOI^X@HP|F-=uBchj)VPTD4_ilYXf3!%pOODObqB>Uhla6WAWe8fAoOc#>e zZ@MZK`L3&p+Q!0_uP+z5|9(MfwofPag|HgUBA&SpRQBzU#4g7vz56`m4?ssV7agR- zobmxk_4fGO1itZS-czf8m+RG58m%10!Q1!0T}C=f?!E;i7;)D~5O_J2ykWLYDrRAu zu`qJ3&iVk2R{qD&RG+B(=e?#dIutoc$e^~kXIB|5bZTPkA0R+D$X{6`V~KhFry3SgbCx-hy|q4a1? z+XIy04O9!xDRwvx*y&>urztT zBDK|;1)QmU?v{iNUD5v_Q)^@A()IR2nPUMhhJx=IIaWr!5ulq|@(T2_?qQCg8-w<= z+qa4hDI~_pIyFZ>xT)LA?F(LF4*$>_Y#q_Q3U%d1D(y&o8)$pRH;1>`p}FOO+Maa0 z?~&`(*CRISGmWq|)lb9DbB;!gBIC*~d#!lOd4}*6rblMB7~!hSWSmL}z!Pe{bY)gN=S|UHpPiU~L)@GwrxM-L!GGq5UWI3H^D4dbYW+T!>80`dU}F zy7hWg%j=UbIa*`sj!GKrjY5*uu2GDJEGFjk3$7feI<6`zq5N_sHG&+IqEJQ3X1e9w zaS7$-AMD4R&eYV79~6T3ga3J|C##@i#mMQUKgFQY{Ik&0PtJNNPZV~4&(-7{PbmLA zisutJSg-aU2mbGim?z~^=8Xt3eKDkx1So=m4y)3Bng`D_GhK<7LY~fb-97_An6n^( zX_9f-rnp4Kzwom^(P|wia!RCTu_1g(g2+4A?}21T1K|Zz;dwm_^ekPOh!Ll?<-Or;qG>z z&?D>REq>U#+DXMpcj;P%+$D=rAxcArDY?f3~s2d~+?nnPH+VJQ;!gXW(Ed(x<4 znjj1<<=Xd5o+*9tB&g*l+~b+4$m!6=Ne8^jC!GFj4#)uqixO-xrG&7g6oRKhv*NT4 zB(QJP6pX9ZP#HYF6iLYOCDUlcaP4y5w&X|Cg?58u4u0Ye*4(3r-a~G)vk~vTq*WVu zb6V4*_gqI+tfimr?Bjot%;Inc<8&@k>#=y|n@D`ISb}Ta*kdIiAmnB-;KJ)aE^CTp z9lS9K4iYlE?*`#se~OZR;tS$zQt~03&p)S3Zc_wo`3Mo!<%*^-u&J8ayB?L=@m&&d z?4bI%`BYG3fy@2PU3RE+TqKG$GDo{FDjq0r-nqOfv0RJwk;6JBGrfI)Vig7nG~zR* zA8#S$IN3>N-0bkm$^Z;l(qT2*=itx=Y3=o0&u{$p=P?~8-S-1R-}w5Q1V#!yEXzYr z{QCLwi*tH(+xm{VedgX{xR_2ag)KnPf?#CTTctVY@sg@SQK?xXvMZA+9>9w?Q?R)# z@j!@{%HT>sS{U|J_O+-LEU~D!W-@co<80H`BwdAxMY64M*f1~QT~5?3dnF{KJkrxy zOkAfxR|1^AsoAsR-3Nr+qiq|*JHwlZ5hYnG#I zwQo;u;!DNj^iGqZlKt&tqR}vwe9Kn|t~|zHSpD-nLxEjBWJv8h4x==wiYrWa8(RjzK3LJdixcb*dc|sw-$>~K z;&Jgd0eT)Gr+o=BwD|6gLd*&jnW1ddyz`h`zmHCabMi`n%e!tIK*@#M+G~1;!Axrs z0c?KXwD-Iflh=PrTkbz@i`Gs8$mmrl!^TSHk47;`#*~KD1;WYkH%l4>AIz4{YY+KG z;?d_`R<{$3Rb->%0SF;5S^b}?0GH?)0uE8qdjAkpmPG~(kJfC`fSHC}@%lP}yjIWe zyW#_z0heT1s2Dp?(kyp@`;^$|m&TM?LSVy6g&z&Q>vX7u&w#hbG`;aM?te4=PX^*u zcqPv1d14Mcl_(3n?N$F+gms%_SLz`fAR(_1Oq|})vT~eR8qo#V)5c&$m6=y zdCP3CmZq#{ErL3m>B-D0>9X&t-=KXJD3;Hev~k;+_`8QT|D6iU7*-jS&t3n12~cga z`GsqP%CU0)QXLLjZ&l5jxw|*5!n|azi zR=^h3&Y(MhB7grW9ebmpk;(6^EqVI+!ywQLPapb9T@R7p>?>%2*9JiVqXEf8a-L3~ z83b@`LYM4?v{#MJ*`|Ogk_Az+%dlzRa9il>Ab9|bkOquO7sF7N^?Gs&bomKgi2>%@ zGNR0e-?h~i2Ww*DklT3Cb)i^0aQaGPiV-2utRN!Cw%*(a=Jw>FuYUh9?XtU6 z4{tJ{hFlD^@aB`IsGN||vFN&7nrZJzRZGM%3L%Yvr&isFs(jp2jc2>+F6hR>(tT~r zC#jwsncTzg0=tk}rH$l^YU7sa^8}8`DFG(>>X>z1+0J<$nHhY`ycr`Egfb}!Sr){{;7Jmsvc%!3#D60L}@JZkT zKV0`l#$PeNO#`UpkYBs~>CBMa`9Mods!j$iQ`qv5yh(lVQKe1bimnK?u!9+n^D(4L zABc(6^s_`*}X35-RqXUA(Z%*KoAc2tlzDMPgH(O?9e*UB6{WhZi$SY8I*@Gfn|Cax< z0^m0oUw9^4RAKKD316tdD+~|Va9@-sFKW$iXjm8%QRF#`Lnn8Y0muAJ|Yn0xpOC{^fCu=qB%;|kILp$c!qHk`ZPw)8Wg<3cQz?S#* z6Xpvm!zJ{vR@n)I`2WN+*$wMfK*t)dNWTSNvR*zH)vf0(s=r3oUa*3S#L;MKIJRs1 zRfs6f=q36C{L;LRSBQl_7wA>$Q?6`V;}C0=!AO>SZ!LF2Uv0h0e9!)`q~Ny=`rlGu z)3+J?<%R>w(;EYN?-P_9!7u^;(FJ+~v_+m}r<{eZ0Gj35@Q6lqxn+z^y55+4A0Trd zEN^)F3d2pAW~qQFPi{mofBZM`|3g5?dwXyRjk!m+*>)+vHt)vv3q{0ho4qLgX;gir z0%nJYk4C>Ty8}(977{%3H|%+~{aoh3al^#J}fu4lE>;|m*I$p~;F8VN3JaAXGMH#p9e-!>huxXd%*6ihD{1J8R8bvYW9 z>U)BJr{7<%=|)k{F97z)yYqLY_Tg_01Cxet8wVOy?>GQFH2{YIe#6BfTFZ_GlUk-{ z)15D5XdkEJp1Wl4c%%9qr>eVd?ffKMePf5CXRlxZ*(w^DNB(-IV1WB!%EKqNzxHsf zVS6hv_|5ieC)^=_^EeBE=5k?FP~~9tn|?S3JBPENAV- zo2|Z|4xib3-T6#1-yGTIr-K!s)jMuZJ-UBsaFjTGecJv9#i>gcx9M zJ|%?Zl!I#zRgPpD-8X@VuC6J@TX1Wb*|jf5GPB`hRfp#+sMKXtdWQ3W?lZ-0L%qR% zr49$=G24$PM51?L*u*2S_SltvsiqkTZ$5q`MHAW+smO;zeCrG8{gwM!nOWzP%L5Xo zc^7Myh11ms40>F?VUTU487)m~_{k5r*;nxv!*_5m4)3k0WY=uEP5{uYH8s&nVeg3D zt;3_MQxya$woCLiG@D9Vt=?9p$6-$-61rCEV_urywePCv&^2{&*Kf+$SyBJCkrqOf ztf{g)65-V~5@*4*K~U^pBpzB z`Rz^C{5vm?t3!wGR^GY%)Nw#}>`g#tacEO$JbAhSVW$rnrROzy#iVtAXCJr?W+ z6y~3Pp4Dx_U6CDJSTPxQqM+1gKF@G? zn>MD-psGXKiHkW9#8BO+o-(37F6?|CnSf~1KH0F06E+v+Q-Zn`7uj}U#5odQab^hm zaUHkA%0n7E;q=X0`}WKHLaS9$Tu5nbeok5c$XX7}Mr?;Tn?Xap-plLe_f>K7wni}9 zHp$U4L&{y$8XSkEZS@kHOF`XmAKs^;G;v3}9eq=EQFF{&H-iz!lpKU}W@U8fJv$!% zft81@b$uo5L0U#0s9EvZ$4197j-Vpirx&Y#b<__<_VVI)k=v05~d>{z~5`jTI3G0R5>o&3^aPHXkCw+o%jqZ^U$B83+1g#( zO^0#eQ?I&Rvg<_k4c(VcC4!MT_jyW4wXxRfcZt$)9{8P`no?B9+_-SLBw*wdo-LXV z_XMv%b@%zBn8<6brkzE=x!+?Izc2@5)FgUrQceh2ihQpUTbA$~H_*1#no=3}e{&=} aF169PZsxgLzEgIR@9OO7RJ7yp_kRIGvsK#w literal 0 HcmV?d00001 diff --git a/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-1.png b/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-1.png new file mode 100644 index 0000000000000000000000000000000000000000..29e262d621a1a0a9f1af7cae768c3754064043d3 GIT binary patch literal 5075 zcmd^Ddo)zv+aH|a7-1$MLkuaETV*0dPVR*!q>@XaF%h{Zx9TWDE-5LuPPq+zg6$6;S z+AuUn29971f<9)){BbyuhSfjMtFufu98OHi+z7G_&X~#+4$ZC>H`&$W7BuXo0^m;**X2O8Bz_1Y|Gxog zO)wd!RJoSwy*t?UznQ=C=dXH-^ToSJ?XkfjqA}s=8=2xWY}vH{$6!xgGyv%l4^GQN zG|?Arp=QK(zTc+6S6m(MtNd_HBqDtg`9@*!Q<%D74G%aj0(;~BZe~Pm*3^c(4bWu@ zQvghp12{NnE)tQYFd}9NAv+tF3uTw*k6iY7457TS4m4ioitX+9gLwHetXBPWRKGMf5gi$!M#ra^)g zD!LlQ#VF*L>+o5~2+AGE7ilvo5vVxvc|?h;A9%ap0`%a4{BImXJ%{UD@5Ulkey8nW?Z>VoL{sEBfrn5MwNF zEHEV9RDXZ9?~nB)GBoC*`q4&Ur*Vze1B3GVVa~2eSJ4`sho|=$iM1E8ogmr(6HH6V zBy?+Dr$3!ELZeUQ!x972H=O^BzU42oqe*bwN2+NtqZ9ZI$if&Sv0ZBmOgFYh z=eNI;kC0yr*O&-~dEFnJ8JT0-hjBCN9ZnltdBP9lJY&-le@DCKyKz$VIcAi}cNet@?ksI2&Z?Dq<2{2;bM0SApVsT zzK&co8)|hoM9)=$u_s8G#jbuWIL^5igUoy?M&gE}jQ;qju^Dc3O@#M5m-G^J2;V|l z<8L6N1%TJD&FN!{un8Z;fzT!j!(o0Kql*-hcIVkf@4BoVB&uwvIQz%!PtIJ$vPlFU zEPQ8$S1lG$Jw6jswP7Q}AxMQavu(`W8vQd!Dmfd;ywiF?0-A{1xQy@1zLyzBuXgi> zIBvh{LcC*=43{THzR#Z5Hk18KqLORH*?!yi>PraR!{%`oo5xFUd^|7wCK4%yQ0Al^ zt?QCBk-Du`+%{-1$_ia#C7gR}A=n+-&fcMS9pNt2h(dt^-d9x_3nY2giI^uyA)gk7 zVLQK#+4fA>)YYkepg7;tW{d~i_Ra|(C{=wgZ0cx#MSJhbs{2l&zXuLDKN#hwbUfoi z*t(ZFjhJ5q-Z5v7rI*^~$F(^xrs&)f`z1I?yuEJUbJ*_SB%EM4zt{W_6x}m0ug%tU ztR0;1E^1*spEBGa+A^qb5xHrYg^#-(isGPY&9h!{iAd{>r%2TmuML$+p~Qgi&*t4b z_xh+bN^K~RuoI%hAXWb8{todWB2`UPojGCYUur)Cl2fB4spsl2>AsfQJa`gQ#z^d& zV=g2hzWD}dFY?y5;{A<}wf7L?s?npgbeMZ^;6vyzTO@f{o^_^COe8&*vpgeN?yXd` zAwv+-8k)h^X})WnU~AnjoiIkL_dG#5&KYSqm%e%^2@TJFhwbw3PC}&el3OHVIWuu~ z^(e``7KH@ro|Y>LZc~X-XHM;m>&cc1)jCCCTITmW(i5_`jSgvzK!+;lNCD*&6<|ou zTnj;`DztB1+Xc6%1K#*E>PPtmg+fO5f;tXzPzBI>NF8n*5d`#ZNd{-pAKJv)ui_cJ zUP^rSto@fg3oo&iB98PeZk}@mX%7v53YcHlX^-?qTA0W%qPL)G{~>YK{!+OK@H|r38EbDF&uh zIRWuPu!%@C;(J~MY94*xEtpuu>973cb%HpZv;b?LJ=?sfb-*(FrzLuS!os7mnQ`F? zernTuIma&%y`s@P2AfR;XXLcqVsOyJXlU#){rg~dsDvWp!fExU%pg7}uCvPu&7IWT zry#g@8SkZ5cyPycNqDo#U(mG7eEUBjugFaPxX$(ByE9AJQL|&M`419>)zNCSngtid zbq#^xpTb)F$7-nr|wto=ISwbNP3eEqJE~yAFwUmK2epYAn z#2%!n|HXw3KCzfOI&O@3hJc4Bi+QDlT7>`!NTvbuC#j>Yssqa)KUqI#|K^PgiG}z#A^z ze~8<6XWZu^?!bns{Otq%a#^ zDvb2ceW@3i5|lDUN62vx2LjI@P>Fb2hW8WF2`M+!jiW#t)Fd9uyg5X!y=Rn(Z8654 zV#HfAE@Pr@^2_(#VA^eW32N6%1(2LLCQcptXGKYZ+V@hS?%*eOM3iUyDSov-#M_U? zzN1T8vN(LI|>H>F8Rz93PEvZsh9a zRlIp?nW`Kvfp;wPbC3|-9xBh0Det|K^&u-#fu)mm46WH(@39ht6rA^f+B)KpQUg>G zd$}q^syRj}1Xx<19jKYlQfO0U#8QPq)r!RfHgAm1+2Cx7Qi$uIT!;pw@Ixe9GGI)@ zN(CsFBvC%*vh;$jb9d3u{S&{klu$@n_nQ4NLgUGwSlsmY4?uEN|0Tp&nNjOa)3kv{ zm4BX#oh~CDBwzOQmhR9CHGg{y)ZEm~3@zPa?qt@?i8)Vr}XPfah$-)_@S;QuQ3N{+pPY#D0b?;Pjmw5KKv?~@ybN|b3x2+4 z2wO4@LUd5loR%u+q~Id4KDF>}_s4Iy?3J5-KK=0M0OTsqff5#O<`KKBj>E_{3R4_7 z%mWosm;}%R4;?n3DT9T)usuW*SK(%wxaoQc!i5(RJEAQfsQ9n`BaiWYI+u8=;*NSB zDBs!33#9NsUKHjU(1Q;;Wk8b#xA9@+7EOxS&JVxYE@AsX?u<0imLI6NgzTg+u9zgV`+SgB4(ac@3#-n&6iMtQZqeexh#1rd(WK3RI+^j))rsnN&%qwT zs+8+S(Whz|XYwq~t5<8Vl>SIqKjtkYt*R~%iIhaS75!MGtZ)-BHUFCcKKx&oo@uQB z-p6Glq*d^oFZqta)>Iy_X2bs1TSj)`penvNsf$R@ zy_?sp)%b`3x;K*#y*Y)0y`!yn?ti?-0G*=kH^ByZ6vK0qvc)VCp$Nk;6(oH4_drpC zeWoQbvHbAv?;O)zX{S?Lf{umPOS|?PC2Pl9qd%(obHv$U;k9F}1G6JE;i{fZWsY_O z(;qz!jviLHvgk3fHp|P`jAGcnvJioId z!97N;SSssY_ET*SLAL1yZkVS{xBi{sX7{Na1GP5AUxjmuGkwSW?K?J=9ka_12^e}b zJ9T4g!bjT(;omv(DFy#W+5Y$M;on)v>F-&`JsNzcgn)Qq*aQ!`Mk0qG^pivQ6PaJW zOjAvjR~pMlG`d~1EB&;?e0M>l-7r#G0H`Y6to B)#CsF literal 0 HcmV?d00001 diff --git a/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-2.png b/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-2.png new file mode 100644 index 0000000000000000000000000000000000000000..0cde719d2591daa3adaf615c716bf382ec0b66a4 GIT binary patch literal 4506 zcmb_g30xCNmyZ%8f*e6a5P@(95D;ZR!`X6ra32v6&_N?|s)QqQt&NC^KmZXY5C|fN za;ZR2&KVUjU?d)(5E11DiID`5aOG>9{dV^^hkg*oKb?0f=MoqPGdEdYq29Swl4phJnFwFc=?(!N4#$7={SL zK}3bPP#*)st+Uq)4ud0NAU+WyCZqsCao@!_91I5up+cnMAQ2TJJ`n?h&_o=Jh{F(x zI3ke>{fJa15fU+(L?)k#gHdr9Dh>zfP>C2S5l5v$^c})vQXv4JN#*mIL=2ONV^Sgd zt_Mk=!iR_t4FQS2qt<~CBRx%(2Q6sxSv${g7)>XLRaLXcQx*Gpe3cf7|VCuT-Y;#pT#mJxA0r z#w1ySPQwO?j`>vqSpG%;*Z~(jBO!Gga3XH{`<uRa$4ar#DTtnJVnZcXPwY;bFCg$~o@m?hYp6Tn-q?-Jf{9&A?FURAkf)lP(w+ z|1I-i)2^D=r)k3nPc?#@ zLKmz}`UaOXf>uLJZpM>xk%jDlZ=4H2b-e9Mm;F&0CrajPnVyQbxMb*sF61v;%)s@#8j;adI`n^(A-8zRNQS(-w z&3?;zMHZ02Uw^6*ztAt@WG*jT0vn$vN}kGbse2!KjMcVsrOn$2Ov?!V+7!zl73^0n z8%POnpjl~VckVg1cGl^t;$>_~@bOj6ewo}H^%Mf&ab_X)Ysj(J)8sHS%{0l*=79E{ z1}Qh0sju6ZYa;!9sgx1bU&NL^M^SG@dOJ3wOU|>oc|DQRyFUjDPuV`#TfLt|DFM%y zST8QuEi@S{!KWxE~crsEx<^6Zn_EoJg(?V z#vV^#B+Gc&q`&7;+BTZI9lFU)*U^)ICQ8Q4Ndd$4PWw^IZOM`mTCtO zMD`A`cZFzIcP~>8(i)iLPD#M|p!3jswo9q;0U?oJDCjQrf~iWQJiAN2$ION3>Du*q zjv-Ov7L1Gs4%pfL#uZ-oyPUa5`WO5;?=Ra|3ulH7E`1qj;oUmg?$_$ra5t0uvRk|a zWS!dAk~K#dIJtz&5Rd3hWGS=~aF_(_3e^}+b&QS-e+5S4k{ZBe7ydGgHlj4rReRPJkWAL*NY5dBg z>yYadRO;+b1NZv#5eVVs$(ms#{?#NqQF(GbP$W@rfTgU~FSO$jhI`eFp4%b|6qy==)zaJ3rzCdBd#fE3zDB)pG-R*!8sN+I>Q#dF%WSfu zgrR|03e=ptT2J{z^q585we3_X*Dim*33S+@J6m(}Kq2Zo$OdxOYSPN@n+o~zOJCa+ z&i3rPmQL?Ftr#gNDXZn2kW5ys-R53SBHI&4PhTK46OqsI@Fok*Mo0VjvM0 zI!y8v;imoyFT}wDeenq)peHw(n@H&+1PFp&6Ght)f#hTlRC$Dg*TvtN3z^AxQ2}~{ za$)fQYzmKkl7MgPdZL8Tx{+)t;f5AZMwL4$hyhOXD^LH&jRB0834mgI%;k>=`uGmX zFH3_|WBi_61S#Rb>ebV9rGO?&jYK+vWcof4`9{lJ{<`jLg*7Ku9A4k$C{(*YV0Vm# zQ|$`=p0{e!7ipl1^SY>%Qc`xdwTrVP3PzF?RqWw9rq{TtpJS3Khyk&~Ku`no+h&Ce z@z;-v?06tP;1hoxZvuuIebwgg7LdHnUXl4EX?|-(24En^I_*LLzn1O-NxhGbiiwm> zJbf2e#j?Y z7~6J6DFjI$Qvz-OPGT^TdJCZL1H(B24J#uX&~XI{aXdv?OPi4K}Wdqv^%pH)3(x6-~H zQ2~smN($l_Q{zp=cPRP^)(jfdN}^1JLB?H)#iG!>+A$pqC(h|^Y|R@~8fCO6r$!9S zWGnf}!`1Y%!Xi|`%!U^Y7)gNUJ>e~Xl`>kztxB_U;`H%6Y?aYGGY;aAFD9KbTEhKv z!v?(kf;WNwTIbKH;y~|RZ8AMDBi2d<%nZvmLcTEJ8VkFc#1C;Jfe z$EAt3v1S3v~9cuOYV8fWXSnxt#)*N{w zw%tVq$a61S6j2@ugG@R+fxOX-{~L{E$+_Z7t2JNjc%QJg@mjpQHu{h!RC?43{?@Sn zWmPJC4DcQ8$!b8}5F%ulyMC~8Qn0jKJ4VMl>m^@2a@q`@4LFsW$ zcl~Lj1DCqQSX^xFW-K2b?%9qZJ@!G=_#fb+3XXmhklpX~(HD3A~AY>H}8< zrl8cOUlT6x=vq7@1*VnVagG!QI$ys*8~@}CoY?V$DSniqY;j@qD)Z7SxME&jYaD%^ zp zXL`s|#o$_LmG^Yz=~|&>hi&%t)4eD`(fI71;>uaD6pJ4DjgB31?g3Le9WY~gfxQ(< z!i`?XOxsP_j~R3DrC&(1ssnI~J$IyN(NHqp1W(~BQ zsJX0NUT7uzdS1<}{o!fn3-zp0*AkJKC%&Rb8M+H3jF)Oiq~K31&&iUhTdpSv!~OTL zmQru{p-oLsqwdk`xQ@Tooi5@EcnXobem7$}p=wYkv+-Sd4|cZC)Z2F=J1;7b>6W*? zIQx(JpO0_yTB#Lhg+<_WJiqOlD-6q@>u7W>DfOBQ%mp8xwD`9l!vF2d_a`MgGgp}x zFp%^gl!=Xiy#*uuB1Jt|P{)k0J6?aeYsqqFrw~YTLXcu3gL31oa2yM5uA{8HA|mjB z?&P$Q_?pVS)z0HW2(Rt%pNdR`QZs{A7HXdsNkej_fXVmm<|8lnS}|~_^39JZQ`x4x zL}Z=Px^8)b3%PrOq6GE{0`};13UrQ}XnV}kEu6M)uF;d;_L&g^vt5+-- zx>KW5(IXpyCM9xWwKUmeG~NvrU~5%fQ8tn)@`L>GFVeujPAJ#JljpY}2NPL4*8gKU MXzgHCec(9w4;Y~yU;qFB literal 0 HcmV?d00001 diff --git a/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-3.png b/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-17-3.png new file mode 100644 index 0000000000000000000000000000000000000000..893c8f2171a44b31615c0a02b33bafcac4354557 GIT binary patch literal 4495 zcmcgw2~<-_x=xcABC<*Z*+f7QZ3WR7BnSlA1OcpM0GX-l11OtH(AP@lrgMnZO5DW=I z06_(*U>^e^Y>+n!0Ye~RKspHo6I1|22|w5b0)zlE!9t=EKo%8*fP{g7(j)?eM8J?p z1QLk~J|rrW1hSY+5>r4WK&S)^l|TS>s3Z)PM4(bZ{19O>si1&>Nfii~Bn*>8U{XQ+ z&;vPO5r7bYCP4NNsSQDpkdnM>1zgbPt0(NQK_KF+Tne*$A0^6fxyabI~H*geeebQCsv>5Aq@wtCIBZK??EuwP+4)=X5O8Q7~hR%#NH}f0;XMnvr59yUqNY z1g&RY zF59Wabf>dg?$H*VzVj(reif1FrPssED~9}FbrXTmqymc8`{z^d zIxkL4cRyb^bKYz(Ie)kBOPE9(-j2?A@z?0FRXs%Ef3qoP2HQj;hOh7 zDcUk)Zmezpn}NA8re5x-Duzd|)RG7cf8Y9S;GjpFudZI3I%9KdQNhB);PlQ*wt=&_ z_OgtID|yFgFVJz@n470Qo7T5nb%xc^$rsoKf&vYt?wiXa0ekjZ3v7bT=digfltLeC zbfJ^=SnvE!#$HS1WlCpsyTFD+*gfJIwld1=5JYG&+uqw%?$9wbwKsiSeZSOPr;RYW zHm>b#CKJ7QBO-5Vwc&b1Ce6@#uKrQrCrZS4dz76oUP;H1)3BDvwNA71^$YTVNI zK%#sI-r)zDqc2WV1;?r2iai3dV%AsmMhcRt*~_?n8N8g4h>gZraMLrQRDhG;?~F4ut1h>oOKx)>kIHmUe38A9|CW{v^YYzSzKj z91y;=HW4+Z5ZBe18=aE@oYU&pTzC>NKiZ*OF^?*6RjUb&*Yd;5>1b9|`wd73RROCj zo?kBG$IQbY4T|D*Ems#7&26i%Nftklns;lDze2&5%5t)dDCh6VG>#RAI3(*a$DEV1 z;9)C~=xW@q*nk|LT`hRt2w3&xosCm$ns;K7AOO^C}zUYqyfEXBlMUa&785XBEwW2UG) zmYXbr_u<-CYaPb8O8e(f_tC$6t_xG_ahzju-(EJos9AV%IL*-Vp^{V>RZTiWc2pr} z;GAE*%H8Fwd^FLY-ukTZ*U71Eh`{_015Oe3&g%s?_g1p2tx&wVR?>>{I z=WEpYP>u?jW?PY4MlTPtdkK`xPW=94tgZW2_Nrz_fPZkqYzf*)5y*eHi_v>5lHmNv z&^um$z6es1<&fQpwMG;>1uama+38rJzap|5f$T1WTJO?Yn?%-V5Hg!pYrnB}bw*id7B5@68+GA#S45)>HX1{YD9*j|aWJNArSusZ6)D)lGEu|s<6w_WkvuCc( zvj%jwQnS&<&C-Ek#L`1!w}--$EQ2p!vJ8!nLkZnP6=3^9*6*ynl+gh4RkE=6iDS|B z$OaVtuj8%-g~nxIq~35kRCWKIt~vnKUruVFC!Qo1Pi!TQN-Ob`!KNDS#Gm1%#O5j( zNJMF0_~ySE8KhXeG2!YDK*jV>iCcOfkd+6EHyec4W%B$ed!x7*O(4I*hwR_w7W`X( zPV;VDPmO5xE7_mnKSo*i{ngmnO5xy!P!O3Zmx^_%TtpMEhxI=w3hnn&V9%{mxhGS; zY}FXYD+_lt^yuj>u!dDO`hyMkzIJv6&6v!b)oQN-$2{dU?md&fRmfD01>Uf;GvJ9= zdMO%@bA(zsMbWaOcc)^do}3{kER^s#>!TCR-?t%ZNe%h$)w0vv z`%cUR!@s1SW@Nu=VTUOOzG_a^j{PMABd{bNbYFOSS&)2BmS7J~4h^LHH905_W~t&( zu3L#QDIdVl&<>-qN|npiaH!uzMLfl5R z$j@m!maaKXsEasNvHTe>n%`z_K*%z!n}1!)bb&Yu(OLTqBi^p}%|C<1*J+OIKg29E zpzXnl=DkK2hu@Dl_GRx5V|KuUy}`=E%qy8HKIT@JO% zXV2&+LYp7sduqf`x-(C#kK%Log$u^y$dE63TOGgLhp#d+Q{3?lFYz#gk-Nyr5W3Tn%+WI z@qV?AimRCp;FhWb=jx9yF!bgT$fO4_{EA4Iab^5g4S5&%O;o85BSOSY0qK&+{hXka zrnu&f0P>}hWBy5Q{y&*}@Tn00Wr-kB{NlS#>GgqXjdHnEk43x0g}pggn@s@s#Jp3I z!jDSH{-d+VUt4m(IOD~6b2BmqHQjqxis<)RSdF2NsIL7KBgNhU8)VMBLIg?(P5LZf zg4GEFTy(@iAyy1FT)|*FFl9uhizz2!mv%CSZ%LSu!zKT$p2V&YH$RMrN`6tcCx;Fm!p`X z=T~4b_ifvlt&aUZ0IoflP-}2k7Qi3P!_p2UW8>-~*{jmTj4j4-84gl4*LfZnrVrf6fBChaBzgm)qbi8C^n2xK?PC7Y0AR2Y|sqMqH{N=&gnhEo97A1=k#_BA9+v3lu##}ICDR@a@`w`Jgkun6ovjx9{zJq{+HT%crsSaT%>wsdxzFn?wJ>nXJD<9DFM7+ z5vfe&QLa@|a3$fTp1~0A{>8mQ+3)^IRU$2hECT+R^gsBolB_Ne9;;<>t*UT27qJJW%D%bUF8A?#-}&xtStcvN;nzXfeovd)5)|q!4bWSw2z7 zQR?$d1XdfEpK?A6A8oImw|N!Vaqq3o6=<2AwjBeXq>U5wVTP=`C56Wm=%Euw9Gqmg z=s1^5;#9njv|@Gqk)}{@AYH(b=4D{!@0RuLpB?ezd}s=Lyw7-3S?^J3W6RDz()+&@ zDSuyB`YF%aW#JdBl_b#_(E|gxH?daU&;t$E5AsJjUK3jWk9G(hv4xM<(~h*%tG>@> zw5~ar6~$q#kU)h2ZOIv!Y=G;jGpL9jlOTn_PdQ&|M|LNl1AB7Bt)|}ur`$VZu{0H& zv?+PuFpVd1m}n03%t!R(mT2yy8c7TJBw(%V+!?(?*`4w9WnWN_WAp0J{I{?>^Jktu z5=4@vSvCkj+=E2etV|cqy*Wz;e98KMpjrU!@@cqTs9Xeo QaezQ9Os!2Sj$gX*KYQOPs{jB1 literal 0 HcmV?d00001 diff --git a/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-19-1.png b/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-19-1.png new file mode 100644 index 0000000000000000000000000000000000000000..4a2f2cc44ebc390102077438c62e187f818621ef GIT binary patch literal 4518 zcmcIocUV(d(~luEMFBxN#Dai|AVq_r1tcI#3ko72unS8-rHFJ4U|?;460uN1xrz!a zO$37w0TlyM0zn8ZBq%k4Bx05lTHcFWzWwKWp8dY(-FqiFb7y{YX3m^4_qjJPjy792 zD{Y2AAX{yZTb+PFpa2B2>7$qsC~-KUD*`sH7<(5q*g_yk2m}j(2p~u#1c`+p2@otu zBv1-IBO%xg@y3NkVhKo4P5_Ak8h}vjcQtlHL&AbW5=a695<);CK_~$WAz+aN0+v7^ zfj5D~Ab?s1gTN4wun-a!Ny1`D1SE-oC6Pc=P){N;7$i_DV2}g?1_8++U>PKkz8is> zjZ44~2ta#K{T;Od1O?i@gK%&t;uns)heIISnl^r$T0)C|hCrl}ZLKU^VsdAh@qSgO zRIh~%l^xCtCkN+Bo~Jt$QE@pn6zbW}dE;x1T9PU|f4M(SXi87m)jySf_`IZ4PKk+n z?Xl;=zw8Z-7rlH;O!mWni#qqyv~6|Z?F=H<+#zGUYLP!{xQ}igJb=4&N1ba;POo4L zYxXJY^9~A@^G2?SVOG zOwulyr|+U=w`HXI+)FRB`!-FST}YnkH3cdRmzs3%jF0)4^UFf!{ScgVAN5QwGh`&5 z>*f5mk$L+qWr}}0X=c(GL8q^*DUP398@$;VHvnhlwsvP{eww(Z>^35cU;cqPa;70n zuxkX4Tfn5B*Y=<8+(&c0xK=hCPL<=qj?{N<~$xNhcg74BKhYU$rvUeQh8j`0Yz#YO~X{7cTa4 zU3|kn_!Of)j)^EA2^A1q=FH&Xo*nG7EAe_X;pv98_%PW@^JH9!$1n@-O;!JKay@}8 zCGW%+i89A5HWirgZBxRikL_8C!Z{O<`NK+HMk@=oN+KBFQ_6ll7i$PwRU1({H&k^{`j^#w3`_`#Nb6r`AaGE#vN-TA^&R!0!3BUwZD>o?{h-by>9a}7*M3EOYKMFB%6PoV5W7Ecm zW2a;S*WQyOyuXYtrdyQ217e@+3Q;q>JmIT7uJVabg{Ycw{vzw1XnBo=9KgP;c;Pmt&a_OqJPp7^MI$_lY$e6)S`+GFLhp{3YL zyLqSMk-3z9JV&wuyZsPv(jC{3Yh%t`mHg7DtKH}xVnMXa2Avz0IpSAuEF@UHsS0M>yNmOQtN^RpZZ#B(x7Gu(Dd`vNJuKn|Dtin`SV5JYS@SYWsid+)G<7 zt{Q&n8Y;Ih|JLWYy`K?f@IHh&ZsOb5irz1*@X@H(g5{^n3#p2C$(&|S7bSzENZ$0l zM~y6Xc%r*nLY;YKS7S-nXhnnV%rX4ri6@*hP}Mj1%=cmXP*n-M$4=TCsbb;ZS^OIy zYlxYFR#5d!_=;hPAMK({o1WIFLLMd4ZFIJw(*VEwx-;O{wJjVC5U{F&ZSE$%X|IG$ zS2xSiA3B_qcLD3R zyg#hEdKAA~Ow*QN4g8CQm1Xa@l_RdGe_mDagrcR0C|=3CiVSqZ(JN+|V;Y$1Y%m$o z?pS$A&w_JbH^S7wmuH4_>QQ$8$ZbA?0+BC{yO&=r~~DNsMMmY+|Kr9Fr5mf&INb7x&p+VdiPvq+U2R~M($2w z6453s?If)ri(S-dh3es7E=r^rSqIF`JT_0|4kT>Xg3g&}J(B*{NKv@14mWHzqHY^z z%lSSka1So8j{n5%nHIv@0D%VL5%sSkZ|i&`;;c;{DhS89c^$`>;C7~T>#%}NPm+M^ z0W$2kSpMlC%c}A!bug)HBQy6#V+r(|BM&Yr8V(Hb zYwa|3L144Y{6?gG#x!bt!o;m9;`=5H6isLV*Ts61<mSs)yQ4}NtzDRjT;nalQcf3%*72Eds+Q;! zA^h&4U-3^`D{C0Fg)UHZ=^ZF_*rJ$$Qk%o^^zge+2vBSL;@)qW<3yyOq>C11<|&G} zabtAdeK1tx*%wP+QDkqay%^YMvM_a&1duhq7IGvyo5eM4OK5&HR{LH|bzX#AL4l?! zxFQ?DgfV{YO0M%-doX3OHuFHVSxPI(>Qyg4lY^S6$+)mFeX$44KEkhp@+_FrS4r@N z#$q7JGKb2T{V;vl@s@>j%a=7^UAuuda!^_;>Fc2V|9{6HDM0;t={`93PZZhbg~TH^ z*uMy?rbTyD{c~z_oPo0$OI^=i)P5Lt5!yjEKKY|eVBFj@FI{=YXBg5E2^%7|?qQo3 zCWI%hSplKX?z-#6!b(bTrR4=;5vz|oQa;J`x*z!|fE8Y>OT99~h4ZINaZMU)VSB1T{0lbuZ5t(|(^+&&$tQ$gVKw=y!3UQbo!$2Djy zv4|`@13Wl#?YU<)pn2O#fT`$JM-{|{uQxQ?|2RbAhF*WX5#UqS&5f%6EnrTS(PbbO zv6Q=EDqVG{i6iDBi&xsJ@)?zcx+IyF#e~)Eq1~1)?nXh&Zt{kO@e9hwfx}Q&mB0TN z%dwU3+XE?JqO;I4#oezF5-zi+FQaydI_{#~eyEyIRw!cK5dvjW&1s)X_F^b_*id5k z1DTes-yIltp~lp3V4pGh_*;f8#rU7h+FmqjsE#Dg9Tr!qae%f^;#JoZ(s_~F8@aTH z14#?Y#iYobx36h8#y0xh+-L=Tp{#%m8-9I#(BV_2ks}~sVgzPO>|BXGa8e1>z+Sj& z(DrSW$TLbB+AqVtD*~p)BKV$o#1BHpG;aZ0T{(+1Q#?v2~_W|s9PjQ((-tp{@ z?03?=r?OE+hX1@f&XHkPvnNQ*7f#1{&^}M(Ra=XTXbLkg7WDtJolxLxPRIL32Y7?7 zIr`&&$9fGUwf%wXx$MTznnqQU4JyFEqLn%O;U`tKqM|1c|1XiCz=cv0;L|dS!p;sI%XmQg-^yMSJpPz9bHhA#(>|tp`tN+i1 zrQQ{QPb{_d7y2Aa_1%(E^IAvke|Fnn+wL!hrH$G7URih<+n&{b;&k>;8gd@ZukOEX zlQ5{H=}*3pJ%!D>apk^V#p7rCP4?Q=X|V^d+*8E*KJf22zpu38kj$2Ye%^LJ`CrXG z9jM6j*nGgCnKaadb4?&lZir{8eDd!U%lmouXcZ6IG2-{BkEVDPo}&pO$!5v zE33MJ{;z@=SW5Ca@^$rAdEX7BeLtH&*|~f@kl;}^ocQr^jn_xy=@Aj z(~=iTB`y2gXEYZ}4(U|K?3{HgKApCvR2w|6knY%!ReUcS4NN>wtLgC*SstDIX!mR9 z?KC%RG$&!St;0{Ao9U|JJg68c@AJ&FuykN7r^3E(pt9hlKF`SZTf2Tj`PsBPk7g5q zQ|Iq4Unq6c2v&>9f2m&;lhZ1AXfsG*RA9!qug~55@Gi6?&FzT?7*faBjpfm3lQ!Fo zwaxbe=B}Rq4R?~-AK@RBdc$3jmx=7r^PV0q&rq*4g_a>U^&N20X})mtgg~){?{T`^ z-cL{ZKob$Y^CoK7f0Z0czjetrL&SPv_-$8E4pMB4SnoR>+VUMy=FQX!8~(F(_gxs7 zNPr(u^Mqpa^Q3&W(pce%fU7767kqi*+(&aLBT-@@1+}spBSZD0X^*sU&CFHmrHXGL zX@<5sf={IUefa$EaJfqhtL8x#PF8@%Q`8#$bH=oS;a*yUSg%$1mOAgr{;E;!bwk)W z-wE9}7iTJX&-d$6F7g+hb!rg@V8|z7Ivz-{y)~*3QD5EgWNlT7ef;^)zodSmO8r@= h{{AxZ&-o$2rctho>qpb_wvF#CwrEGIa?2l+{sX4WU-ke1 literal 0 HcmV?d00001 diff --git a/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-19-2.png b/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-19-2.png new file mode 100644 index 0000000000000000000000000000000000000000..568acb2f03df70006027d19ffcb25c2678f788bb GIT binary patch literal 6357 zcmeHMX;f3$mX1$G0kMDrlt~c;g<=p;2m(eCL=ccs8AUXR%*8x}Xix+}h=L#@5=3SL z$PkbrK>-;oLNE-H5F$Y#fB`8cAtZS>sp{(W-dgpiSND6>y*fEL_uhT>-e(WrIs4w5 zRC`-Xv0eLifj}TJtFz_~Akg1XAkdCaLOTJe8_XC_D!BC)Kz(}Bx0U3ZB%Hx0{kx(QC3J@_EC`NH7Qq z1sa$;g2Vt{B!-Lx=wu9%%)vlG7$_JJz#xB<00zkzfX=~SI2eilNcQ)=@a- z$FJd>trhIp>-*p5#ySewFjw}r`=|W7_zAIU_7~%X?`@p9)^$Ug4~~isT_IX}u%Mh) zG8CPH%M&y|-Z8lUdcZ>n!3Rgq&Oijeql7kj#rJ+&{pr#z6EMXD0(djC@4~SV3iXuEn&V=9+ z##~wCPd-cT8(-P}9;VU_23z@XcC0GIju*KMyrZb!GE>_u;p<@PWP>2%{Dc)jj~n%H z{vbG;9on_IrF&hCjxm-}uzRRN7XnQ1%R$`_fs;)Jqx}}qh?>Ig%y_~=kM;&wz+~r; z*f+fp@h$RPSLKNACYT-}TAB4Cm+L9iw3~x9`CE|L)&9Wc>dW<}?$&0d z>8AvrOp*F4bf07d*n{}=Cq}A6`K7ycUbeb7jyX`U3$2A#_I{TyHc3Vhy}~1#wjyHJ z&L~3)ZuwAL2#|q+_;&qTzDR-ONOwqKurZ`*Y0b(D0p8i!LP|Pk9}URi&aVz6p74yz zJPW9CZx36YFEV-nuSq;-AF_wip)D#r55VI4odb}9H*IzI@w5ffv0{}t;_!}Z4u|e_ z{csRcup?4lNn&`fea|w$^bZ$+5$u8&&s_1epTd?e3q*yhT;E;$aytP&H=^q3VQ4yY zmLPxGjk(EwD~1mW8#-&Bc-bA6wyO5%e%Kq};-rG>Ug3W6^64W7A)7lQJ+7&!o1rqQ z()qp!4u?H@fZDzZu0AS}tz_ThY=QW7=X`Vj{|o)E4ulSxm5`YsZYG_Y#`WWUzud%{ z2s!M4hL${l1JMOjs-0VKwNm{w4M=`R?PBnnzsCgjSCx>|G)sX0KR|scHpxe((wES0 zAQ8b=+8ABM)?`0omGS-Ld=O%JQo(OuoPZV=dfZcz<<9t2zJVPp{l^mgmL(z{pltek z3mf!tLYtGnJY*QBFAQ#H^lwgYlru~hnE4}5{9ZUd^}k-o%5FE0|3?+WVWmYd{o34r z(wrRGlRGD=)Lh^7qXj}s;-X6;7QU8*3a9leLu7<)0?eq?>~xKLe4Z+`Xl!@c=jnIN z56=?L{B7QlNHE>^(HdAYs0=+ZM)9q7bHohjUr)Xn=F{}{j@^+w+V%rf8sFV+v2Uaw zgJ!y74+8evCiX{3$&8gJ1NIv-W3yTk>rC)J2)CRK-R{J^S|PGG*&_i-DA9pw zM~l<*ZUv(c;}rlV9j&h}uHyXDJjK&*7s-=@>6BS7gurge=D7ewn2*$IDTe=7XolZ{Em`CG3s7mMyri-9a3l_HRBbQnL z-8cRHVfF5wW{WOy_q08fA>kbj4enF(Q#1oCRem<(&ux61^K5C&#qW`X@ROaTC6PIK z+O!F7q^3_XEgaq$!nx)&6R^A8bnr^NczwopL`d_zew?F4F~`gDw9Cq@-umLkwadOI zFL`~a!;7X$7o|ZL25y4!`VE-^>k`@!zwEAo(fl_)uVNqK5OLE!Co0>7w zS<2GS(wea6#_DF+)FwNaDpWa`B(41bWpvr&x_5a9|A^YWH5cAGTIunG$Wh#r&TfNB zNYyPM9YhDXXiH!GYefJH0%!ri5_##?+J&aPdFT|PnweF6ex`jUH0=Hj!j1iq@ROV%<7e_FlzB}oFY}bKVmplml1&MBY0$BX z?K^9lz&yJMFq17otGKHA2rfb>uEK7wxXvtA@u4pQNjbUUjBwEe+(9wHtTZE5@#|mz z2w9#ZZyVsP;hAACfEyIaUFSxPf2?OeTgLwkJ5Yr=7gvt0CAO

on}+7Db?*H?*I? zFGPFPcKw*l%aP~;2C@4_u-E2BUrzH|USw`^i~N)~47*-&!P>*;x&abp)KOc#j*=Kw zkR4^1jV9FFP2}JuTxH$FxTP08NpZpg)t#|*5ZGjRD*nvNUo_$YF1_h-t;;z$&(8KCqp7Xh#iDbe@@l? zXOl0wJX_2v_qE6RsL=$wdXG8^96ZRThI+XoGWrRFqGzX=3qwX(1k?M3##QWWVf2GYAYJN?iF04#h zTFWy^vrvkv!1HhDn08>#o#Wj@!C>L}3p6$Q3hTv}nXP3;SVO)oA};&8aEm)^pG;HD zi9WbNMikM?K2w4L=>x*-{sc9kORp6f(N(AFG^jrFTW)oROLO|!j=&wnY4=}6Vibn2 z854520J!Vj5-3Ol03fYNxA0pI>?Gvqe1#Cljub==Ilhvw8}2_W{rnj3Nlyn2Tkcdo zmCc{?NE?39Jf#k})s&l@mwK+L^OBh?o+flbT6T0q(Wx1@L(MhQR5rf^Y<{%6XKt&D zJeS*}%+KpiyxX&suNm8KC~Nwz&SFT(vGnov^lSH@o0Zz=4#~dXcLr{gcjdWmj=??l zg9#ip6tVi_W52are_{@w9)@v(Zv)J@Wgw- z;lJ3SyjG=~iN(t*g15;8=|*O?&EwD zj{4|;5xKO8@o~5hkcq{HNFOsewARK7DoEoz4X%TxP%77z>iyy6SLwk9o;$lEAYucp523$?|P zwt*F;cvpnR*YZKjaSyonX#(;5`mA$C3f`hZoh}_sbiS%Lqof)SE~R#mr_r8Ml}+Bl zUam?do3HwH>3uq6aZYRH&nID9J;Mq2%*L-U^Hwq+iYXA6dXp*Rec~e)b8tr=2Rlpu z!kS8w|}HvQSv{Pq6u0XT)9 zj!_F(8+#%5(F;2|+06dpe0)u-$Xx8XoKP8V_s8JKM5EfPb{lp^}% z)|vZF=zX2&$tbGns@>9E>eF2$_HhpzEN8izdLZ*X`VDL9P0NoBWd7X`)DX{133tiO#^|-gny?CfpCH^>O^)J7rlS@1nIv2~k`lcCY3E2=CSQ=ab}Q@(v(YF$Fys5X_Nyyv z9)x$;HHuW2GCjr`IZ%P}{VW2pQ=~`d%ibw)%R()s*6fOF6D&}Pi4<>Pu2XTa$I4!d zkd0{;p9Q3>HymQ8BpY41fVGJyMEB*PW_wU$nFa3dHr(SZ!j#JSvgcHca zNV23Y?x$1O8shriEuP=#pQkjv5LhKVzX?ns!ceO%?qU;jcT-b)yumuH_!C(*p*}{Wph(=L%5JKh-IvUnIP|zR{ zmk}&Ph5>^lEXpcEAS$>Jb^-_iGROo7A@DBv{;I$Js`qND{+XMT^u6Et`t<2@PoM4z z73+RXMQOVd1OicUJ??Z00)a9hkWE7hG9ZFEWh@6aZP*j1oxv6YK|vq{2xJX{LP1aj z2#N+FfWib(@EHXmti$UU0Y#vpK%52&59|O^2^(s{x`s&rK_)0`G!%q}LV;8o0YW37 zXfy(i#soh!CXWVcc|00#jY)to2`DCkz@(v=Gy;th`-AEY);bXc4uyQy1*f8T<+yJo1ftrw{@v7ix%eyuqNd~OPh2u#;SzH&`bbBpqyv$e${hmZGIrsb1x7 zv*iUQx|h73{hUgTmFX5ttVq(^w3Bso`)>G@E&2llk@8AZdi9BcZL)NSyJOp1`UA|9 zPF)^p-_boYj}hAr&h`lE!lg-)LEDT)v3XT9`Xo=7su;+a{Cp3FFOJ?-`qp{YnK6dL&7E`LCZ8wH) z@NCt0@X6A7yi9-!jWUR-{$#oLvho&_7x2oA-)gCYVU!hKkvPc3G;rP#1Nz8%Q;07d z|0lZSXnIs`)ZxJJ_xtV*Cj?YSJE zRio6atYy%K80P0NZgA4eUpx13 zua;E=RC`_Cy5GGusy6G7ybo|eo^D-4{F2Xw7~$b??I8-;i!5n|j6G)0ZN1&1cTBzv z=ejJwe4Wc3n=O(cBrEJ?wyIBTs^AuN*y)087!^3p~pe zMaNZSq}u}e3MB0Die7m`pZn9v`WbIM665y_%nED~$e*-iQ43e@90aSB0PUGEqK@ZV zd>QunwRr2LQ+aVvXKmof_6?!GCC#yu@7o`ngQZyJS$r9o`b zt%q?7j%?sa?ubVA`}ogbP}CkGM6U5k_HNc5nOZT%(N}A_smPei zTj$~`6dtS_1_KfC-5Y!4h}MqgvfNc<9bW3nP-u$B9^d`9iJ4MVwD<+N&NXW!tM65# zI$s*EWB+icM;f%Y#OTpWFMG9SB0~6~U_S}lRG}DE zTa$Rm`@Zi-k1b$6^_+j)5j2K>OR@$N;s1t*-aS1I91cdn&%>0?y8vOc(w}Sa@LOp0 z*L%5@EG4VlQ5T(TboKPMaDLnqEdp4r5=f`%h_o4 zoAV5s%b$}EX53ErqaA1KU@1&b=%t$Aq)*~4$X*4FcmG?i%CjVp=FYA_R6IzubniY2 zX0o3UZM7)AZBuYrKlZ4ihU2mv3%3m4{G?756$EwmA4L0j+tSAFTq*NB^K{j`R;}0Q&Aigs53Pk#FP78h2w5 zZDyV?{f=hi)b?OulWaN8$Oi@+vE%`l3mv_d{K)HzJyE#sxjQPQD>FMow|TJ;oZn3M zJR{7L^2N58V)w^}HX&Es%ubJ?Q_-PR3u2eLsnN#bM@^Ek(C_{crBRPtez=q&D#9_! zFvl5CYk05SE+C)K|AXCfD_C{z|%tz3dT1pgitau{@bxbu5% z_)b+|!HF?yuLGT-qFc$c}ALssfgG zDK2)v!aCWU?`8kk1hfc-CxE3~EtlVL?bUm#-R%CAOJ`=1Rby^v?mwYroBwIleh@8R z^QKi}uX2I=ba-Jnqex}n0xyG(ogP*rp7dP0gdhqPgZAtoUR6CdhlsO=ZMRn9cD(@n zROD3}9ZEOqT)_fCox$15MGPK-R^EkOopQ@Z=eaV9>U?2D{I}w^uNJ&FL!~&(aP-mt zuA6^;C!8sTN|CKTbvvco{WM6}8d=M^|5g_l2msV^0Lit=Ypoe?Xi$y}y&vB}GCI^@ z61$cD|AtTVPdoSBZUs8YuZ=p$J%Rj1hNN0@zqXw35!ebZMEKC|IFD-2u_pf@0*}F z9vqG4SspAn&GR*IwMT3)Sx#M6v9tA)V-IH`_9#x?G_ zTKA;IBJfWC#gMHnE5CF`rRAckN(@{tE}{iqKZ>BYD@+3(bW>Ww>ojsUzrTZNYWSU? z{+Ay%5G5;S*G(Z%3rEJ$U04JRr2(}l-(0!McZ?k!5qQ`?rFbiJJMAE@el9&fr+!x~ zTiUWz+^=^(N|nHIn9-#k0* Rz5YjttFyb)vm^ib>0i^N1OWg5 literal 0 HcmV?d00001 diff --git a/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-21-2.png b/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-21-2.png new file mode 100644 index 0000000000000000000000000000000000000000..704b1bab9270410101efefaee402b332c3d78c00 GIT binary patch literal 6107 zcmeHL2UL?;zE6S_kv1bukv^y($RMB)DWQvWQ4tVuU`;{AOZE3>Gy`LKd z0`Z!jGCTtULC_!&c$A9`pjezy<^WEO*5+r8fHMdL1%coo5EBH2f}n5^6a|6<0}D`r zITQqEky!`u2L*)#0}Dj~g9zvV((oM#I0^+vVc`G~i-lu>UnU9)LcxInWpPkg;Df>v zQ2?EYMG=`;I0y@e0t#5vXAOW!!~%3C7RzLIcXty}P$CLW!~z4w;s6H|;4=YDCK1IX z0#d+W5hiYGZ~;l|zIMvq9|YoSWPQOcK7~JnK*BgvL%p*hnbfS^!LFu3Hx1k?6$jmu z5H{M*`8`*CTpHH<6WKY5fKwQb79uNDP@Rbe%ERcC|9%HFf1` zY-}yYlsP>=OM)+-{B5|&}+12)i zh>d#mPE#m3gE8|PiDinBm-%U1N$oltrTC9DjmqzJnfo`p@e8e$cy?Ua+}+w&?fCE{ zewQHKDDp!SEox8(`+Owhb_V%lWE8o4Iiy{yfe6wyfDwN%+7&Rca6HcqW;b!Q>T>_X zW8uThPwS!DSte>=>L(53Epav=79KF&y=z-boIqE>U+$h~&kl|PNi-X5x6x6=DP;@M zg@=`GDwUNkLCpTM^g(~(%%eRs29EJ$9x7R7@pwZg#h;L>3P#*rCoy?C*TLc)hG}%m z3et$;ugFM={fy;oNyi&g3l@jyZfOG>JaxQ_D>Mb%q&aL0zv_|cbfBfRac|7hf%22> z!;J33g!zcz#$3o80+}t1Cb4%CJN599WIH+$)u`h`>Y}eyMTioPcaiBU%6r%Y_yB8V z&??K=syK24eaS7#Vl`jeGy8P{l>2m9+?H5dbLPQt!>wmj44?B2GEha zkf9G$pW2D=;K>o=#np4$Rh;z^3|gT-L3R}IU9LUjQ?Tdh@x+%KdIkR0PCwLhjBsq4 zYQ1JY(q=>8FTIbtvg*G1!>lCjmnkbeZFx?CQ5%IgX$#{=R^Hfnx4e1TYFudX_~KEo z4jpDb@nq!0Q#n_TI^H_7zO71zk!XlmNOm}VjlS52=K^~0{{YCNuNoG{O$$xsmn8O! z33LUOdG?=ia_J;>qF~{Ef`kJCUD!P{MNa;KWt-tRP2Iq|xyUKxZ7c$I%M*if^}58K zENNJdV59kw*#jbkI74^ly%J}^C44eN>y@!MVnUjkND`!#fBeCTef6QKGfC zi9ql5vYFCxtB<-ugfr^gGhBaw(;uMx7hNwe# z2g2u8?myoemPhA#3lhG&BjyIPDsdE%-(jPf&xe)YqATxnA2I>AgD0A+IF)|b;e5Fo ze9ka>-nT3y{X9pv(V4LihtZmUWO_Z-d{^lrWXe1cc{F%8%juBAhz@?3Oii_SacmwG zHs1-v)5m@ID5fw8YWCTeu^7R9uVqA!Jyet1lP?{c8l2HYU4cI`2;cE>X=q6C+DbrB z!orC*C&=G60O0D(HeE7`5(L6+N*UVQ}u&4%aqcAOXkb+3S6i|1sVSS6>rQv)}`9l zZEN${Wk}INw6IkoPjIGSc0F}oAiu;>Q5tvE{k)>i|)+0mHg5lqXMir@$9 z&t+G~DRp2k>HqCj_>LIeV=0spcldMHcy!a~OkaUDfTIXzFUC=Q%gY z-cWdJ1M`P%ge)Q^8?d_Kju8!deDxlNi9h$uIf!?4A%^l=3da)d@w0L2kJSs01rjZs%&Ecb zjo!TL0JdoG!H)3!o^W_6qvQnJj_NzFZ{n&2OL9P)(W&usH65AFX*(!7T10{R?{@(r z0R3+iq;e3z}fzYHdp~1GFrt}#-G}*uS$vsNA z@ip!LCJ(8SkAPULpB#EethG>JTNbio2I=r88SdOh73T#G*$IC3m%mibSblp#Klekc z{+Hko3v24=BJdNs{&t=)^ z*8T3y?}0R}+l&I8JS-U;Zb$bN?XXapw{wJzK=T3)Y$)jhXfyT9`cbBTNt2q^Pc-qO21?D63DLwnB z3V;tmLO>QY*D3F2|9!{~{~1csMQiKNnIGpGt&Bcc8hmx)y1#Uh;8Wa^pjZw18k2Mm zb=@~^a&BGI@#wCv0o?YiwGWp#(SL|y{0;m=O4iy@IV0tKa}^uO{@!?J^ZTwj zI+&*-_36i&48IzRc?P@|{=Hifmk!l)Tz!c96n{>cw48S^e)#~=WvIC17-ecwqDpna zy_aWj+M2@S%y4yZJsQfedDpH&X(n7(Q{QSTN~5}!J%5qjUaSOejoiK3JmtpgsAWcuaGLY3C zsY1zP#=m^r5$2U-cMzNvSmvFSMCa>xCerw{pQI-P1R%#NJ}yY8=v9A@OA88|ow+zCgUiD?y4X(Kc`4aRZZUpZTGbe0*NqOTB5ag?IL(GWYlH z;Tdno60Vo|Kj~jb=$Jmh$BYE{UUL!3b)L(NK8CX)i|=tXqo>TY&bcn3)5bD9EL z$?7TmRdCH(@j3iKQUh(T%PxE?FEv=lOjBA61ZU|iGJ;Z)A}9-tEfTA zaboRD5`$2pJdh>3+POWE<3-k98!p+c88^HgvZ5An?9_TSK z;Sfs%9sl;hm`ZT)dKiK6BY7k_ufCDk7ExwJ?F&NKaX1I!d7?BmI1x<8;Ejp&pqLD! zu*TZ+^D7rN2&ICx%>!=_hd$%fj?596T^~xwV}u7k>r@;NpO~% zl9zwFo6a6kC-=r%f?%_J4(4ICT5ECVMpm&rdLh<_WAz$8G-mXJM8L`{Sy$WQ`_c_` z?NTJf4R+}51Q4ufa8Vo1Jd6+wJYpwx>iJ~XUMW$@iXPFCj zM&JA33!ApO(;GeTQnL_3i`3wr*;a_u&b)qAxZQMaHO(GYOGwOKe#b`pak%T*Ae+R+ zCabMeS|8bH&$gQPwi{ILkx;G1$JxQi{4L-tJU=v3WJF7e7%{h))|doMqh@4v^N#Vc z4e9pc_=m(tGUA`}GkfuD&(bpa#;fQPltayB1GDtPwWZ!nh{DLLw$hOx;tx+9Yn*^B z#cM*v9g)@})X5Pl=iI&qVuR44AM@9(juOOGSlHUXX18~o30BICTT*n3RA+6{n;Kaf JV)QTl@>hAIb7}wp literal 0 HcmV?d00001 diff --git a/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-6-1.png b/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-6-1.png new file mode 100644 index 0000000000000000000000000000000000000000..5d62474f3b9075a2b5c48b8ecb761e09a82a81ea GIT binary patch literal 4231 zcmcIo3pmqzA73+uXt|`MDAyHo%6%>y@ z04RVIct(KGLb32dBhW+yASVJ~0}g;G`l}i()KJiXkODwJM1Y701Yk-;gNSGZk%%S| zDZqzFVG{u@n@wa3C}6e5B`L{lh$E1;)P*lY@*6|gA+0h@?m6VYr6fUiz~ zMtBL>0s-I;sJ~hX4FQ2p;E)Qi6^TfD?`RNcN1O1wtuvzPA_xS#=y1ZyBerOs6MB5S zW1H=71f79$_QkA1zn{rYEV4;+#rc8q5Sh>ofuO1vqwcLBo^A1PNqg2@MFIMq*mxT+ zu=7|b>b~hG80h&8)WKW-CtN;FxYyxhIZ*F4N;B+8T=|uAjoX=8x>ep(CrEdOZ`+rlI#8i16p+c%bew?_w%}3j`r$r&-lrDjGy=aqQ_JWTGo`WO8910v16nN*YYNuJG+12 zQ@_)u$!PdsDDui?z^4E1PH274G`ja)aT7pt4o=XW)DKWl7mj;+hRnzvEH zzG0Gpm79wptz;qNmPb$PGWQEt#!Nf%skp|-hoa{eC(fu)L~thu*n_RfgSEVwl6m7Z z{_|m13wJg2u{DBcd%4AVkROMkEZd>c7f+bTt@)ab?y7sUTdrlsk5Ic8j;@3(h?7kR z;6@J2?lZ>ocw@E)?e(XP0^Hr}0&I%tU^z#)zpHs(+GIh`+u0iAHWl8Cwm`QohPslr zw|N@EvixXN99+0IpTwB#t~j|8N2(Yie|lX~+qmYgRJHN_ZqG5~Gwnhc+gNCDFXs#+ebfhQhA;7h9QSMLfC4d{y{h zsj>J6Gs61(uTgTCDZsA_!ZPKXvf~OuB2~zZLxm~xjVX(F6j{)DF_z9&n8uUZFP|-| z{GVg(tb^+AB(N=DGbvTC3r~>-tzI>nwhbWNMz>=>z=1lvpQXoCnTk|{eV1Z z7%4w0*7$o8@Y!6B2`rhj0GoRYn+$)ekHOy+E@Bfc{yu{2f@?WXOIBX4K+U@6Eg)Ms zbM{74R}qI#u8+mML&5RBBa*cUI&ueEGRDGZjn zT#|Dk@zh-Hf5*vXpD&Z&VoJ(l!8ZN{iddZSnl*#zX-DFnTuGf6+E%URx?I76>e4s7 zXwD?@Rx8($7M@xzG2p9_9DrVklHBmQU5I*}ug(}OrvLoB z!GFW!nwKz4Szc6l{ag_4QFd-Oxr{?|ZOw6P)aERwUw-FL|(76-{}F(7y*CCAe~p~t@F zMgyn^+OwSAVW$!BJ~vs>^_IAj_#>rP;Tc9tF)KRvtz`6^NZp?&jvHRTciheal6l~< zvwq%|s$^ZWFvT#rXn*{?jt4S2_`Vf5XgUw1Tn#eVAPT%;%ngRx%;NYN8yu&h;;0Q_Vg+<}59$JjE zpVv%p4#jJ$vr^=i37hKw08rxSx>!bZX%3+2|_kW!lCKTUDf z{9hRu2yI6d$u|qF{8fZx{bKb-^;?yJ61s7l-8ztA=>GPm@p7+hE?di*$~wy37q2`J zK9kmWEvh+o-^my5#VLwX24g{Vz|`SkR; z^r{LA;V%MEU#)ut$`qfF9d2(R=X{2rXI%Rk_pdPv6|J%=RfLB61GIM%vGh&gZl}pVpCSamoi-(X`lK*$4SKj@|#

Ats6W!n?cm?}-inlAm+Lc~z7xaEQTIZqLJDHlyYgb7k3I24d77WMZmt@{Y`?p>m6q*7dfchSs9KT}jenPA6X>MHc(+uk%p>S2 zWZ010N_rZyuv`=@=?8rUf^AY1f}p$}m~$4gxBmS_aPf@Pda1;w>r;ZQE_KS1X9f5e zK2lJ1vS5h#>&S*$>z+X1ND^`79%%$!si`0?Vv@^2xpTr2k++Cw-taL88D_Rm$CYiiq+ z{*#v2C8B$a+Tdc5URTs4D8$_hD6cHBY=D_AJWx6O+4nqW=+kK9`61<6*T*8d^Bmo} z;fYEY{msC9{z)&V+XS)=@m`Rb2*5BK@j z@o^n#+V6%S=d$z1vypgGU$~~?V^iA#PZG6WF_?iz*4gF^Juka(Sf^hz2E`PYLchve-ljk0!naiSqdxt)4@QLB5)#--` zo+|BHEuk*P^hPaw>B#1l$@qn^fSX-EL0P>)rVHK{9^59+HDimIRn5`1UFo~CP=A*0 z|5dK}G;%THrrH69y7=;0aRUjkg#y?qP!6-vo3YZ-&O8@!*+dLda6+jnVR5K6hzgAM z{J_O#Q=v?b_i^{?mhSUScC4N_)*q1IRmfQO8l8RPf}YR^6VNk(oSXCB$qNuS$k1e5 z>R#9i5VP1x(0mP<(Q^zgNMfX-R&7WQ*kH!Qu&+i|*;?C9jO_N5fc*0*-uW|QoaC90 zi;m1lB`BEgFmw3+IEobKu|evNyEi*jvI|TvH=rK*4)6ERUFyGnvDt!IVr*{szl{7S Q{6oRP#`#38^|z^i1DqeBq5uE@ literal 0 HcmV?d00001 diff --git a/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-6-2.png b/Rnotebook-tscounts_files/figure-markdown_strict/unnamed-chunk-6-2.png new file mode 100644 index 0000000000000000000000000000000000000000..bc43e5847d497f0b01165a50fd18d250a35bd34b GIT binary patch literal 6024 zcmeHLcUV(dn-4Mul>j;_5?ZVX4n;zfC?x`-B25%Ew2xv4XaoZch|;_w$_xagOEG{_ z1r$PDLersXfDnp+gpPp>Rl1>NZxHm^=bL}_nSJ)#XLpm6^kEhz0-*t+Oedc;*a)R^Wzy{NZZpE41!vWIk6d_M-F6NA3J=$nkL^y8j`mW%2O z_GhZ&hD(WJa}SO=8S{ViWqRwxw8dFq^0FY8nX)+RmFN7m*$Y+Odls+%gIYjab- z%}uxkP1y>LB2;jb&xN7I zcQlOhVAPz90f*9+=z&@7dt6F>(;*}fdoTZ`mgBl0Tj{s)saa?bV)i%u%(XB0#k>94 zsOn0cr*1`Ov9%#@CJJXJRy%sQ&esAKV%_F0iOtD1)wtbUh66ahm82A4*!nnbm^`&s zXR9FO>YP?jxY!ubE8jqBq6{c;IhsllHAQHej$j=f4kkV_cT9WaKDz$wo!V=iuW6@F znP<__zqK|@=v>}9^U2#efJSQG@>LlS-FFJ1Be1&QmoV=(T-D-M)4W=;#9@$1(kquo z>_m~c-JmgNFZbk@fn)xQ6Ln22wdN2kqvf##d~O@h4+a>n7HJ6-iZ;5_BIz5Xkq6*vDHvMF*c`zru6k*&#Mw zC5NloB<65};goHW;2n;|*TC3v(jaXXx4TH7 zja5C44#C%(J~>afuJM*TT78{yeQ2WoMX6CTZTszAo?X7ngaN*h9&oYUlX)*}P9x@I z?C!xWB*mjxk*xa1pZG@!4hwg)Na)X^j*j( zp;Z?;!IxEU%nv6EkQ6n9oEXyo%mi-0a;!3;pcpvwH8?-VloT+FON1hEB0Qt-Nm>xE z5|{nJ0;G+V5d`bv|2xbvRn>`of$MQtXQsL=X0O}hQMF^%Lr#8}(UZCEI=_`cUcR)R zfS35wU!T#|12NHSowQg_r^5x;<#!AWE1A+;5xpp)v#NTw1U;~>4MJjX3JuZzQKXzE zBsj&h9VpDaW#?*@KiTISCn46muE|4lN-mo2<++(F~FFB3lNcLfLT-X~W=gMuq#wGTV=%d?F_;SBH zVujtL_oS$wderTPG&#O?1igcR$@il3(CA&Ou;-PndfuhuRB6$D>x>5z0=!X9WN*aPqS3s?4bE2ZVfGDPVokC&uRW>~*udrn}D% zT>iC&PE=uIg9Q67jb6&%RW_()EE{~~+E_336;Mhx!|GsB;x z#%>Yl+GBfN<2u70r(noL!P^cQK#!P$X`h6G$QzA$=NYN$zN4U!sm%TdIW`@i3>%F+ z2t480PynzYoR!>z9Rb39^4?;efAZsbf=Kqv+QjO#{hRUmVwR->IWG#WbrHHOzgBY9 z+8p5}ocu@b_%Mvx(2z|;g_wYzETE0y6NxpeOl+#&%3EG~^`ykDleS9v)2jo7kRJ&O zpT42l)&7AbEr*o7^Uc+Q0m3W%322c5qmHv0qJD_W0mF#H^8$IB>dHXIv%t33)`xzi zT%i#kPE)(D=tnYM52O6+&XRDJFz}u5abIj54+0iYsfX=^Z<#om&F~6_WsL6+;8MoD zi?S~Z;>VoKh|%v9pG?|b;k-#8yH~BJK_fBaGHtMCehAMm--^6)#_yJyU%Eg0+tL+k zv#@l%Ty_LbMave$F+1s&n zda8v_OwSYa7=Nkus3a=*N&?Jvqi9iRg^g+zXru$Xk366 zlZ+5J(8&jq=r@oc;cXjLmjN&vnEsyDU%9tX!EuU*y+&u3Hn$;3=&2IXW9_kE9_w52 zzKXZKEJ4jqMOoE1Q%s=qXQ$P&R&$#?`o!^@z*bpD&7WXPot~{yoW4rkG-mClvR5Rg zN%3=2o=yuUoZ^MdG7OzuNBNJ0cB`{X<_$3$*6ribi(DD<{N-=$mqT1B`=*Zzw z$!IsRu02h;UtJSwi-;Dko9Qf~Km4tte&5soiXHlYV+H?dRR7c(|;1xEGD#+Oh7mVGv4{CRhzZ~ukaB=z7Qozj?A7OJBH zK7a9Ug=}qJ-~ZShb$cM+b8nM648Nla7W#B9*CBSU(DIqnfm-PL6S#?$-Adu3omKZ0 zCG6JLC&Ens{&bTA54TEDR^P@MJP;H0?6g$Lau={suUNUh3uGiyD!#5yzCXhU)`>av z&LNwQOT!S3m~?g{Eb4-4E%{7yB3vc`_q$72nuf{+WV2YSxPcos-YHEvLoKwo;rIuA z#|t<1lqc9fd0ZRcF6*mOb$EA}a-R7d_Tf1h@eM=r)t!Ii5TW*$a_FS3RRH`(v;(Kb zT$H65TPf*`kEuOAMSNSuPJdY#tCdNae79`ZX4GI)&Q*?o>e!w@@;Z_;U4J$Grd=Vz z(s8`aku!E%lP5rHX2Ysm;}J_(TjCu%_Gvf z$TrzhhsQa}u&$QE*L@8OEPD-F_@}*5*cTObphl>;YB%z!7iI_XxH*DYNafnY0ClT{ zk@TcpQent6bcavQmB@+LT;=B(dNgQ(NU7EzId2q#iZ?5>S|=M|<3mQBQKvQRv%h(;u`e$p6}J!D%xDJ=QP zk?VLC9r6?0Q>$IA#XtI=H&Pcn{b5VvTI#rqwOphAlPIUzo6nv*#|(iu$Tozw@|!M!aT3 zGJ5)q-);GZBQaOfa$hToikx{3Ooi`jFe?W)%l+t zuKvppn6^1z%V&EQ&UF3RlH_&R?vCPe!PUT9J)Jq$QS_^Jwl@m8)@Cwtr(@01z_$BC zu7i&$A0EeVBv);Fc^3fDmCU}^?5%=_XL*F;FqYR zGi<8(>bgiDf;-tPoLf&ovrvUP)gz`rDL3`1-K!IZvUZ$01B%~ti`8oxr^}xw>q@Uk zRG*Kn_(grSs#(w=a;CVj(-vbsREd~XNSV2kF>a$TxXoDyhz0-Grpj}6_x2)+G>fk? z)$HItmscDj0T0GXrPuVSaNlFy@-qUxELKWxStj8{r&GJRh9HSg)@VZ?p=(`K+ke-DsyH!G6|>i z{pnKHDbn&DfT#s2=B5~#bNbqI%<~g{X|Y(PNpAd?J)tUJHj;`yzbDm=kQUv7-?V2np{M%7 zg{XcUqhAO&8B70$H{F2*+K&*0y6&aYC=LQ6BqaOZ2_$azm_I%