TSstudio 0.1.3

Converting time series object to a prophet format

The ts_to_prophet function converting ts, xts and zoo objects into prophet input format (i.e., data frame with two columns – ds for date and y for the series values). For instance, convertig the USgas series to a prophet object:

data("USgas")

ts_info(USgas)
##  The USgas series is a ts object with 1 variable and 225 observations
##  Frequency: 12 
##  Start time: 2000 1 
##  End time: 2018 9
USgas_prophet <- ts_to_prophet(USgas)

head(USgas)
## [1] 2510.5 2330.7 2050.6 1783.3 1632.9 1513.1
head(USgas_prophet)
##           ds      y
## 1 2000-01-01 2510.5
## 2 2000-02-01 2330.7
## 3 2000-03-01 2050.6
## 4 2000-04-01 1783.3
## 5 2000-05-01 1632.9
## 6 2000-06-01 1513.1

In the case of a ts object, where the index is not a date object, the function extracts the time component from the first observation and use it along with the frequency of the series to estimate the date column of the prophet data frame. For instance, in the case of a monthly series, where the time object provides only the year and the month, by default the day component of the date object will be set to 1. Alternatively, if known, you can set the date of the first observation with the start argument. For example, if the USgas series is being captured during the mid of the month (or every 15th of the month):

USgas_prophet <- ts_to_prophet(USgas, start = as.Date("2000-01-15"))

head(USgas_prophet)
##           ds      y
## 1 2000-01-15 2510.5
## 2 2000-02-15 2330.7
## 3 2000-03-15 2050.6
## 4 2000-04-15 1783.3
## 5 2000-05-15 1632.9
## 6 2000-06-15 1513.1

Similarly, the function can handle xts and zoo objects:

data("EURO_Brent")
ts_info(EURO_Brent)
##  The EURO_Brent series is a zoo object with 1 variable and 378 observations
##  Frequency: monthly 
##  Start time: May 1987 
##  End time: Oct 2018
head(EURO_Brent)
## May 1987 Jun 1987 Jul 1987 Aug 1987 Sep 1987 Oct 1987 
##    18.58    18.86    19.86    18.98    18.31    18.76
ts_to_prophet(EURO_Brent) %>% head()
##           ds     y
## 1 1987-05-01 18.58
## 2 1987-06-01 18.86
## 3 1987-07-01 19.86
## 4 1987-08-01 18.98
## 5 1987-09-01 18.31
## 6 1987-10-01 18.76

Lags plots of two series

The second function, ccf_plot, provides an interactive and intuitive visualization of the cross-correlation between two time series, by plotting a series against another series (and its lags) and calculating the correlation between the two with the ccf function. For instance, let’s use the function to plot the relationship between the unemployment rate and the total vehicle sales in the US:

data("USUnRate")

ts_info(USUnRate)
##  The USUnRate series is a ts object with 1 variable and 850 observations
##  Frequency: 12 
##  Start time: 1948 1 
##  End time: 2018 10
data("USVSales")

ts_info(USVSales)
##  The USVSales series is a ts object with 1 variable and 514 observations
##  Frequency: 12 
##  Start time: 1976 1 
##  End time: 2018 10
ccf_plot(x = USVSales, y = USUnRate)