plot-aggrbench.R

library(ggplot2)
library(ggthemes)
library(dplyr)
library(directlabels)

our_theme <- theme_few(base_size = 24) + 
theme(axis.title.y=element_text(vjust=0.9), 
  axis.title.x=element_text(vjust=-0.1),
  axis.ticks.x=element_blank(),
  text=element_text(family="serif"), legend.title = element_blank(),
  legend.position = c(0.60, 0.70),
    legend.justification = c("left", "bottom"),
    legend.box.just = "left",
    legend.margin = margin(0, 0, 0, 0), legend.background= element_blank())

aggr_results <- read.table("aggr-bench.tsv", sep="\t", col.names=c("nrow", "ngrp", "sys", "median_time_seconds"))

snames=c(pandas="Pandas", duckdb="DuckDB", polars="Polars", arrow="Arrow")

aggr_results <- aggr_results |> 
  filter(nrow < 1e9)  |> 
  mutate(sys=snames[sys])

print(aggr_results)

breaks <- unique(aggr_results$ngrp)
labels <- parse(text=sprintf("10^%d", log10(breaks)))


svg("aggr-bench-groups.svg", height=5, width=8)
print(ggplot(
  aggr_results %>% filter(nrow==1e8), 
  aes(x=ngrp, y=median_time_seconds, colour=sys, shape=sys)) + 
  geom_line(size=1.2) + 
  geom_point(size=3) + 
  our_theme + 
  ylab("Time (s)") + 
  xlab("# Groups") + 
  #  ggtitle("Varying Group Count, 100M Rows") +
  scale_x_log10(limits=c(1e3, 1e9), breaks=breaks, labels=labels) + 
  scale_y_log10() + 
  scale_colour_discrete(guide = 'none') +
  scale_shape_discrete(guide = 'none') +
  geom_dl(aes(label = sys), method = list("last.bumpup", cex=1.5, hjust = -.1, fontfamily='serif')))

dev.off()


breaks <- unique(aggr_results$nrow)
labels <- parse(text=sprintf("10^%d", log10(breaks)))


svg("aggr-bench-rows-fewgroups.svg", height=5, width=8)
print(ggplot(aggr_results %>% 
  filter(ngrp==1e3), 
  aes(x=nrow, y=median_time_seconds, color=sys, shape=sys)) + 
  geom_line(size=1.2) + 
  geom_point(size=3) + 
  our_theme + 
  ylab("Time (s)") + 
  xlab("# Rows") + 
 # ggtitle("Varying Row Count, ~1000 Groups") +
  scale_x_log10(limits=c(1e6, 2*1e8), breaks=breaks, labels=labels) + 
  scale_y_log10() +
  scale_colour_discrete(guide = 'none') +
  scale_shape_discrete(guide = 'none') +
  geom_dl(aes(label = sys), method = list("last.bumpup", cex=1.5, hjust = -.1, fontfamily='serif')))

dev.off()

breaks <- unique(aggr_results$nrow)
labels <- parse(text=sprintf("10^%d", log10(breaks)))


svg("aggr-bench-rows-manygroups.svg", height=5, width=8)
print(ggplot(aggr_results %>% 
  filter(ngrp==nrow), 
  aes(x=nrow, y=median_time_seconds, color=sys, shape=sys)) + 
  geom_line(size=1.2) + 
  geom_point(size=3) + 
  our_theme + 
  ylab("Time (s)") + 
  xlab("# Rows") + 
 # ggtitle("Varying Row Count, # Groups = # Rows") +
  scale_x_log10(limits=c(1e6, 2*1e8), breaks=breaks, labels=labels) + 
  scale_y_log10() +
  scale_colour_discrete(guide = 'none') +
  scale_shape_discrete(guide = 'none') +
  geom_dl(aes(label = sys), method = list("last.bumpup", cex=1.5, hjust = -.1, fontfamily='serif')))

dev.off()

# just explain difference between n and nlogn
n <- 1000 *  1:10
nlogn <- rbind(data.frame(n=n, O="O(n)", C=n), data.frame(n=n, O="O(nlogn)", C=n*log(n)))

svg("aggr-bench-nlogn.svg", height=5, width=8)
print(ggplot(nlogn,
  aes(x=n, y=C, color=O, shape=O)) + 
  geom_line(size=1.2) + 
  geom_point(size=3) + 
  our_theme + 
  ylab("O") + 
  xlab("n") + 
  scale_x_continuous(limits=c(1000, 11500), breaks=c(1000, 10000)) + 
  scale_colour_discrete(guide = 'none') +
  scale_shape_discrete(guide = 'none') +
  geom_dl(aes(label = O), method = list("last.bumpup", cex=1.5, hjust = -.1, fontfamily='serif')))

dev.off()