如何检测常数周期的开始和结束?

时间:2017-07-28 10:46:52

标签: r time-series

我试图在R中绘制衰退阴影期。考虑以下示例,其中衰退期被识别为1,非衰退期为0。

Date           Recession  
1918-09-01     1  
1918-10-01     1  
1918-11-01     1  
1918-12-01     1  
1919-01-01     1  
1919-02-01     1  
1919-03-01     1  
1919-04-01     0  
1919-05-01     0  
1919-06-01     0  
1919-07-01     0  
1919-08-01     0  
1919-09-01     0  
1919-10-01     0  
1919-11-01     0  
1919-12-01     0  
1920-01-01     0  
1920-02-01     1  
1920-03-01     1  
1920-04-01     1  
1920-05-01     1  

任何人都可以帮助我了解经济衰退期的开始日期和结束日期吗?预期产出:

Start                 End
1918-09-01     1919-03-01
1920-02-01     1920-05-01
几年前已经问过{p> Similar question,但我认为答案是无法解决这个问题。

1 个答案:

答案 0 :(得分:2)

data <- read.csv('recession.csv')

# Add new first and last row to the data, to enable the detection of change
# in the beginning and in the end of the time series.
data <- rbind(c(NA, ifelse(data$Recession[1] == 1, yes = 0, no = NA)),
              data,
              c(NA, ifelse(data$Recession[length(data$Recession)] == 1, 0, NA)))

# Create new data frame containing the start and end of each recession period.
# The periods are detected via built in diff function which calculates 
# differences between all consecutive values in a vector.
recession.periods <- data.frame(
    Start = data$Date[which(diff(data$Recession, lag = 1) == 1) + 1],
    End = data$Date[which(diff(data$Recession, lag = 1) == -1)])