Error by changing all the date format in R-CodePudding

I'm trying to change the format from all the dates in one column. I tried it with strftime, as.Date and with parse_date, all giving me some sort of issue. The column includes 2200 different times currently expressed in the following format: Feb-03-2022, it should be expressed as: "%B %d %Y", how could I modify all dates?

ethereum <- read_csv('ethereum_2022-01-04_2022-02-03.csv')
head(ethereum)
# Changing date format in the dataset
ethereum$Date <- parse_date(ethereum$`Date`, "%d-%b-%y")
head(ethereum$Date)
# Naming the datatype and the timeseries 
ds<- ethereum$Date
y<- ethereum$`Close`
df<- data.frame(ds,y)
View(df)

When I try with this code, I get the following error:

Warning: 2200 parsing failures.
row col           expected      actual
  1  -- date like %d-%b-%y Feb-03-2022
  2  -- date like %d-%b-%y Feb-02-2022
  3  -- date like %d-%b-%y Feb-01-2022
  4  -- date like %d-%b-%y Jan-31-2022
  5  -- date like %d-%b-%y Jan-30-2022
... ... .................. ...........
See problems(...) for more details.

CodePudding user response：

I assume you are having a packages problem.

Try this example:

library(parsedate)

## Calling the function
parse_date("Feb-03-2022")

## Specifying the package to avoid masked functions
parsedate::parse_date("Feb-03-2022")

In your code would look like this:

ethereum$Date <- parse_date(ethereum$Date)

The package "parsedate", is used to parse from any date format. I don't know if all the 2.2k are in the same format, so my suggestion is to use that. In case you are 100% sure, you can use many other parsing date functions. You can even write one yourself using string processing techniques.

CodePudding user response：

You can convert the vector to the date format and then apply any desired formatting.

vec_str <- c("Feb-03-2022", "Feb-02-2022", "Jan-31-2022", "Jan-30-2022")
vec_dates <- as.Date(x = vec_str, format = "%b-%d-%Y")
vec_dates_str <- format(vec_dates, "%B %d %y")
vec_dates_str
# [1] "February 03 22" "February 02 22" "January 31 22"  "January 30 22"

For convenience of applying in data frame you can wrap this behaviour in a function:

my_date_transform <- function(x,date_in_format = "%b-%d-%Y",
                              date_out_format = "%B-%d-%Y") {
    x_dates <- as.Date(x = x, format = date_in_format)
    format(x = vec_dates, date_out_format)
  }

my_date_transform(x = vec_str)

Example

sample_data <- data.frame(original_date_str = vec_str)
sample_data$new_date_format <- my_date_transform(sample_data$original_date_str)
sample_data
# >> sample_data
# original_date_str  new_date_format
# 1       Feb-03-2022 February-03-2022
# 2       Feb-02-2022 February-02-2022
# 3       Jan-31-2022  January-31-2022
# 4       Jan-30-2022  January-30-2022

You can then apply your function to a data frame