CODEWITHSUNDEEP
r
EDS
EDS

Reputation: 2195

Dates Mysteriously Converted to Numeric Data Once Put Into Data Frame

I am using an API to get some data. To get the data, I use:

library(httr)
data=GET(url, add_headers(Authorization=token))
mydata=content(data)$data

In a gross oversimplification, I then format all the data like so:

day=unlist(lapply(mydata,'[[', 1))
price=as.numeric(lapply(mydata, '[[',2))
fulldf=as.data.frame(cbind(day,price))

With str(fulldf) I see that each column is factor data despite using as.numeric. Documentation for ?factor says "To transform a factor f to approximately its original numeric values, as.numeric(levels(f))[f] is recommended..." So I use that as follows:

day=as.Date(levels(fulldf$day))[fulldf$day]
price=as.numeric(levels(fulldf$price))[fulldf$price]
fulldf=as.data.frame(cbind(day,price))

What is strange to me is that str(day) shows a date vector as expected (format is "yyyy-mm-dd"), but str(fulldf$day) shows a numeric vector. What am I doing wrong here? Is it something in an earlier step with wrapping lapply in as.Date or is it the as.data.frame that is causing problems?

Upvotes: 0

Views: 53

Answers (2)

SmokeyShakers
SmokeyShakers

Reputation: 3412

price and date are vectors when you call cbind, so they become a matrix first. Matrices can't be of type Date. They must be a primitive data type.

You could call the dataframe cbind method directly:

cbind.data.frame(day,price)

Or simpler, per @jay.sf:

data.frame(day, price)

Upvotes: 3

Ronak Shah
Ronak Shah

Reputation: 388972

You can try this : 

fulldf <- data.frame(day = sapply(mydata,'[[', 1), price = sapply(mydata, '[[',2))
fulldf$day <- as.Date(fulldf$day)

Upvotes: 1

Related Questions