In titaniumtroop/aRlegislation: A Dataset of Arkansas Legislation from 2001-2019

knitr::opts_chunk$set(
  echo = TRUE,
  cache = TRUE
)

packages <- c("tidyverse", "ggplot2", "ggridges", "gganimate", "gifski")
lapply(packages, require, character.only = TRUE)
rm(packages)

options( scipen = 10 ) # print full numbers, not scientific notation
options(future.globals.maxSize = 3*1024*1024^2) # 3GB max, 500MB caused errors
# Sys.setenv("R_MAX_VSIZE"= 8e9) # has to be set by .Renviron

# Load topic distributions from storage
dir.files <- list.files("../data-raw/results-scores", "Rdata", full.names = TRUE, ignore.case = TRUE)
dir.file.info <- file.info(dir.files)
dir.file.info$name <- dir.files

newest.file <- dir.file.info %>%
  filter(size > 0 & grepl("SVM_results_short", name)) %>%
  arrange(mtime)

newest.file$iter <- seq_len(NROW(newest.file)) - 1

# %>%
#   top_n(1, mtime) %>%
#   select(name)

SVM_results_agg <- tibble()

for (i in 1:NROW(newest.file)) {
  load(newest.file$name[[i]])
  SVM_results$iter <- newest.file$iter[[i]]
  SVM_results$weight <- NULL
  SVM_results_agg <- bind_rows(SVM_results_agg, SVM_results)
  rm(SVM_results)
}

SVM_results_filtered <- SVM_results_agg %>%
  filter(!is.na(F1) & F1 >=0)
  # filter(F1 > .1)

height = 4
width = 6
max.iter = max(SVM_results_agg$iter)

min.F1 <- min(SVM_results_filtered$F1) - 0.05
max.F1 <- max(SVM_results_filtered$F1) * 1.05

SVM_results_filtered %>%
  ggplot(aes(F1)) +
    theme_ridges() +
    geom_histogram() +
    labs(
      title = "Distribution of F1 scores"
    ) +
    scale_x_continuous(limits = c(min.F1, max.F1)) +
    transition_states(iter, transition_length = 2, state_length = 1)

anim_save(paste0("../data-raw/results-scores/images/SVM_short_histogram.gif"), height = height, width = height)

min.topics <- min(SVM_results_filtered$topics)
max.topics <- max(SVM_results_filtered$topics)

ap <- SVM_results_filtered %>%
  ggplot(aes(topics, F1, color = kernel, shape = type)) +
    geom_jitter() +
    scale_y_continuous(limits = c(0,1)) +
    scale_x_continuous(limits= c(min.topics, max.topics), expand = expansion(mult = 0.05)) +
    theme_ridges() +
    scale_color_brewer(palette = "PuOr") +
    labs(title = "Number of Topics") +
    theme(
      legend.position = "right"
    ) +
    transition_states(iter, transition_length = 2, state_length = 1) +
    shadow_mark(1, alpha = 1 / max.iter)

animate(ap, 100, 10)

anim_save(paste0("../data-raw/results-scores/images/SVM_short_topics.gif"), height = height, width = width)

SVM_results_filtered %>%
  ggplot(aes(kernel, F1, color = kernel, shape = type)) +
    geom_jitter() +
    scale_y_continuous(limits = c(0,1)) +
    theme_ridges() +
    scale_color_brewer(palette = "PuOr") +
    labs(title = "Kernel") +
    theme(
      legend.position = "right"
    ) +
    transition_states(iter, transition_length = 2, state_length = 1) +
    shadow_mark(1, alpha = 1 / max.iter)

anim_save(paste0("../data-raw/results-scores/images/SVM_short_kernel.gif"), height = height, width = width)

SVM_results_filtered %>%
  ggplot(aes(gamma, F1, color = kernel, shape = type)) +
    geom_jitter() +
    scale_y_continuous(limits = c(0,1)) +
    scale_x_log10() +
    theme_ridges() +
    scale_color_brewer(palette = "PuOr") +
    labs(
      title = "Gamma",
      x = "gamma (log)"
    ) +
    theme(
      legend.position = "right"
    ) +
    transition_states(iter, transition_length = 2, state_length = 1) +
    shadow_mark(1, alpha = 1 / max.iter)

anim_save(paste0("../data-raw/results-scores/images/SVM_short_gamma.gif"), height = height, width = width)

SVM_results_filtered %>%
  ggplot(aes(nu, F1, color = kernel, shape = type)) +
    geom_jitter() +
    scale_y_continuous(limits = c(0,1)) +
    scale_x_continuous(limits = c(min(SVM_results_filtered$nu), max(SVM_results_filtered$nu))) +
    theme_ridges() +
    scale_color_brewer(palette = "PuOr") +
    labs(
      title = "Nu",
      x = "nu"
    ) +
    theme(
      legend.position = "right"
    ) +
    transition_states(iter, transition_length = 2, state_length = 1) +
    shadow_mark(1, alpha = 1 / max.iter)

anim_save(paste0("../data-raw/results-scores/images/SVM_short_nu.gif"), height = height, width = width)

SVM_results_filtered %>%
  ggplot(aes(cost, F1, color = kernel, shape = type)) +
    geom_jitter() +
    scale_y_continuous(limits = c(0,1)) +
    scale_x_log10() +
    theme_ridges() +
    scale_color_brewer(palette = "PuOr") +
    labs(
      title = "Cost",
      x = "cost (log)"
    ) +
    theme(
      legend.position = "right"
    ) +
    transition_states(iter, transition_length = 2, state_length = 1) +
    shadow_mark(1, alpha = 1 / max.iter)

anim_save(paste0("../data-raw/results-scores/images/SVM_short_cost.gif"), height = height, width = width)

SVM_results_filtered %>%
  ggplot(aes(cross, F1, color = kernel, shape = type)) +
    geom_jitter() +
    scale_y_continuous(limits = c(0,1)) +
    theme_ridges() +
    scale_color_brewer(palette = "PuOr") +
    labs(
      title = "Cross"
    ) +
    theme(
      legend.position = "right"
    ) +
    transition_states(iter, transition_length = 2, state_length = 1) +
    shadow_mark(1, alpha = 1 / max.iter)

anim_save(paste0("../data-raw/results-scores/images/SVM_short_cross.gif"), height = height, width = width)

SVM_results_filtered %>%
  ggplot(aes(type, F1, color = kernel, shape = type)) +
    geom_jitter() +
    scale_y_continuous(limits = c(0,1)) +
    theme_ridges() +
    scale_color_brewer(palette = "PuOr") +
    labs(
      title = "Type"
    ) +
    theme(
      legend.position = "right"
    ) +
    transition_states(iter, transition_length = 2, state_length = 1) +
    shadow_mark(1, alpha = 1 / max.iter)

anim_save(paste0("../data-raw/results-scores/images/SVM_short_type.gif"), height = height, width = width)

SVM_results_filtered %>%
  ggplot(aes(degree, F1, color = kernel, shape = type)) +
    geom_jitter() +
    scale_y_continuous(limits = c(0,1)) +
    theme_ridges() +
    scale_color_brewer(palette = "PuOr") +
    labs(
      title = "Degree"
      ) +
    theme(
      legend.position = "right",
      axis.text.x = element_text(angle = 45)
    ) +
    transition_states(iter, transition_length = 2, state_length = 1) +
    shadow_mark(1, alpha = 1 / max.iter)

anim_save(paste0("../data-raw/results-scores/images/SVM_short_degree.gif"), height = height, width = width)

SVM_results_filtered %>%
  ggplot(aes(split_group, F1, color = kernel, shape = type)) +
    geom_jitter() +
    scale_y_continuous(limits = c(0,1)) +
    theme_ridges() +
    scale_color_brewer(palette = "PuOr") +
    labs(
      title = "Which train/test split?"
      ) +
    theme(
      legend.position = "right",
      axis.text.x = element_text(angle = 45)
    ) +
    transition_states(iter, transition_length = 2, state_length = 1) +
    shadow_mark(1, alpha = 1 / max.iter)

anim_save(paste0("../data-raw/results-scores/images/SVM_short_split.gif"), height = height, width = width)

SVM_results_filtered %>%
  ggplot(aes(weight.ratio, F1, color = kernel, shape = type)) +
    geom_jitter() +
    scale_y_continuous(limits = c(0,1)) +
    scale_x_log10(limits = c(min(SVM_results_filtered$weight.ratio), max(SVM_results_filtered$weight.ratio))) +
    theme_ridges() +
    scale_color_brewer(palette = "PuOr") +
    labs(
      title = "Imbalanced Class Weight Ratio",
      x = "weight ratio (log)"
    ) +
    theme(
      legend.position = "right",
      axis.text.x = element_text(angle = 45)
    ) +
    transition_states(iter, transition_length = 2, state_length = 1) +
    shadow_mark(1, alpha = 1 / max.iter)

anim_save(paste0("../data-raw/results-scores/images/SVM_short_weight_ratio.gif"), height = height, width = width)