Typo in data.csv - Oct set to Jan

[ryankenney-dev]

You appear to have entered October (10) as January (01) in data/data.csv. I'm getting these results when I try to filter to the first day of each year:

grep -P '^\d\d\d\d-01-01.*' data/data.csv | head -n 10

1871-01-01,4.44,0.26,0.4,12.46,5.32,109.05,6.39,9.82,0.0
1871-01-01,4.59,0.26,0.4,12.37,5.35,113.6,6.43,9.9,0.0
1872-01-01,4.86,0.26,0.4025,12.65,5.36,117.57,6.37,9.74,0.0
1872-01-01,4.97,0.29,0.425,12.75,5.52,119.33,7.04,10.2,0.0
1873-01-01,5.11,0.3,0.4325,12.94,5.58,120.89,7.16,10.23,0.0
1873-01-01,4.19,0.32,0.455,12.27,5.5,104.5,8.11,11.35,0.0
1874-01-01,4.66,0.33,0.46,12.37,5.47,115.33,8.17,11.38,0.0
1874-01-01,4.53,0.33,0.46,11.61,5.17,119.47,8.7,12.13,0.0
1875-01-01,4.54,0.33,0.4517,11.51,5.07,120.72,8.71,12.01,0.0
1875-01-01,4.3,0.3,0.3767,11.13,4.71,118.25,8.39,10.36,0.0

[chillindylan50]

I can confirm that this issue exists on my end as well. As an example, here is the set of dates for 2021:

2021-12-01, 2021-11-01, 2021-09-01, 2021-08-01, 2021-07-01, 2021-06-01, 2021-05-01, 2021-04-01, 2021-03-01, 2021-02-01, 2021-01-01, 2021-01-01

[chillindylan50]

However, the source at Schiller's website seems to be the root of the typo, not this Git Repo.

[chillindylan50]

Here's a hackey solution if you're processing it in R.

# Correct rows that are duplicate January dates, so the second one is October (10)
# Note: explicitly assumes it'll only find Jan. typos that follow a correct Oct. date
for (row in 2:nrow(s_p)){
  current = s_p[row, 1]
  previous = s_p[row-1, 1]
  if(current == previous){
    proposed = as.Date(current[[1]]) %m+% months(9)
    s_p[row,1] <- proposed
  }
}
s_p <- s_p[order(s_p$Date, decreasing = TRUE), ] # Sort by date

Psuedocode is:

for all rows:
  if date of current row equals previous row's date:
    increment that date by 9 months to shift it from Jan. to Oct.

It will silently fail if Schiller's raw data switches to have a different typo, e.g. Jan put in as 10 instead of 1. A safer way would be to add checks specifically for dual-Jan dates, and make sure that the proposed changes as a whole reduced error between the data in the duplicate-date rows and their neighbors (i.e., the proposed changes smoothed the curves)...