I came across this article in The New York Times today about coronavirus and I liked how the graphs were presented. I know the bar plots can is just using geom_col() in ggplot but I am more interested in the smoothing part. Just like this graph:
They said that "each red line is the seven-day moving average, which smooths out day-to-day anomalies..." How do you do that? I have a dataset that I plan to present it in a similar way.
Thanks!
data.table also has a rolling mean function, frollmean, which can be used for this purpose:
library(data.table)
library(ggplot2)
library(scales)
# create some data
set.seed(1)
DT <- data.table(N = rescale(dnorm(seq(-10, 10, by=.1)) +
runif(201, -.1, .1), c(1, 800)))
# apply rolling mean over 10 data points
DT[, `:=`(rollN = frollmean(N, n = 10, align = "center"), idx = .I)]
ggplot(DT, aes(x=idx, y=N)) +
theme_bw() +
geom_line() + # original data
geom_line(data=DT, aes(x=idx, y=rollN), colour = "red", size = 2) + # rolling mean
geom_histogram(aes(x=idx, weight = N/10), binwidth = 10, inherit.aes = FALSE, fill="red", alpha = .2) # histogram
#> Warning: Removed 9 row(s) containing missing values (geom_path).
Created on 2020-03-19 by the reprex package (v0.3.0)
This takes the 3 period moving average of the 3 points up to and including the current point. The first two point are NA because there are not 3 points and the third is (1+2+3)/3=2 and the fourth is (2+3+4)/3=3 and so on. Omit fill = NA if you don't want the NAs. If you want centered moving averages remove the r at the end of rollmeanr.
library(zoo)
x <- 1:10 # test input
rollmeanr(x, 3, fill = NA)
## [1] NA NA 2 3 4 5 6 7 8 9
To take the averages of 3 or fewer points use rollapplyr with partial=TRUE. Here the first point in the output is just 1 because the average of 1 is 1. The second is (1+2)/2=1.5 and the remaining are as above.
rollapplyr(x, 3, mean, partial = TRUE)
## [1] 1.0 1.5 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0
See ?rollapply for more information.
