bevingtona
commented
10 months ago Current usage logging allows for monitoring processing speeds..
processing time estimate is somewhat linear by watershed area (but depends on amount of cutblocks, roads, etc. in watershed)..
Intercept = 0.306740 minutes
Slope = 0.000257 minutes/km2
Call:
lm(formula = time ~ area_km2, data = dfp)
Residuals:
Min 1Q Median 3Q Max
-0.77953 -0.15724 -0.06444 0.16259 0.87388
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 3.067e-01 3.948e-02 7.77 4.47e-11 ***
area_km2 2.570e-04 1.491e-05 17.24 < 2e-16 ***
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1df <- dbReadTable(conn, "usage")
dfp <- df %>%
as_tibble() %>%
mutate(time = case_when(str_detect(processing_time, "minute") ~ as.numeric(str_replace(processing_time, " minutes elapsed", "")),
str_detect(processing_time, "sec") ~ as.numeric(str_replace(processing_time, " sec elapsed", ""))/60)) %>%
filter(action == "processing")
dfp %>% ggplot(aes(area_km2, time)) +
geom_point() +
geom_smooth(method = "lm") +
theme_bw() + theme(aspect.ratio = 1) +
scale_x_continuous(n.breaks = 10) +
scale_y_continuous(n.breaks = 10) +
labs(x = bquote(Area~(km^2)),
y = "Minutes to Process")
summary(lm(time~area_km2, data = dfp))
Currently logging activity by session id to database... perhaps add https://appsilon.github.io/shiny.telemetry/