如何使用R将DataFrame转换为时间序列（含示例数据集）

Got it, let's break down how to turn your dataset into a usable time series in R, step by step. First, let's clarify your data structure: it appears to have three columns—f1 (row index), f2 (dates in MM/DD/YY format), and a numeric value column (we'll call this value for clarity).

Step 1: Import or Create Your DataFrame #

First, let's get your raw data into a DataFrame. If you don't already have it loaded, you can construct it directly like this:

# Create the DataFrame from your raw data
df <- data.frame(
  f1 = 1:36,
  f2 = c("11/1/16", "12/1/16", "11/2/16", "12/2/16", "11/3/16", "12/3/16",
         "11/4/16", "12/4/16", "11/5/16", "12/5/16", "11/6/16", "12/6/16",
         "11/7/16", "12/7/16", "10/8/16", "11/8/16", "12/8/16", "10/9/16",
         "11/9/16", "12/9/16", "10/10/16", "11/10/16", "12/10/16", "10/11/16",
         "11/11/16", "12/11/16", "10/12/16", "11/12/16", "12/12/16", "1/1/17",
         "2/1/17", "2/1/17", "3/1/17", "3/1/17", "4/1/17", "4/1/17"),
  value = c(0, 0, 56.25, 0, 56.25, 0, 111, 0, 0, 13, 0, 0, 0, 0, 0, 35, 0, 0,
            415.21, 0, 0, 280, 817, 830, 644, 0, 0, 90, 0, 0, 250, 0, 45, 184,
            578.16, NA) # Using NA for the cut-off final value
)

Step 2: Clean and Parse Dates #

Your f2 column is stored as text—we need to convert it to a proper date format so R recognizes it as time data. The lubridate package makes this super easy for MM/DD/YY formats:

# Load lubridate for simplified date handling
library(lubridate)

# Convert the f2 column to a date object
df$date <- mdy(df$f2)

# If you prefer base R (no packages needed), use this instead:
# df$date <- as.Date(df$f2, format = "%m/%d/%y")

Step 3: Handle Duplicate Dates #

Looking at your data, some dates appear multiple times (e.g., 2/1/17 has two entries). We'll aggregate these values (summing them here—adjust to mean if that makes more sense for your data):

# Load dplyr for data manipulation
library(dplyr)

# Aggregate values by date, removing NA values
df_aggregated <- df %>%
  group_by(date) %>%
  summarise(total_value = sum(value, na.rm = TRUE)) %>%
  arrange(date) # Sort by date to keep sequence logical

Step 4: Convert to Time Series #

You have two main options depending on your needs:

Option 1: Traditional ts Object (for regular intervals) #

Use this if your time series has a consistent frequency (e.g., daily, monthly) and you plan to use classic time series methods like ARIMA. First, we'll fill in any missing dates to ensure regularity:

# Create a complete sequence of dates from the first to last date
full_dates <- seq(min(df_aggregated$date), max(df_aggregated$date), by = "day")

# Merge with aggregated data to fill missing dates with 0
full_df <- data.frame(date = full_dates) %>%
  left_join(df_aggregated, by = "date") %>%
  mutate(total_value = replace_na(total_value, 0))

# Create the ts object
ts_data <- ts(
  full_df$total_value,
  start = c(year(full_dates[1]), month(full_dates[1]), day(full_dates[1])),
  frequency = 365 # Daily frequency
)

# View the result
ts_data

Option 2: xts Object (for flexible, date-indexed data) #

Use this if your dates are irregular, or you want to keep explicit date labels (great for financial time series or plotting). The xts package is perfect for this:

# Load the xts package
library(xts)

# Create the xts object with dates as the index
xts_data <- xts(df_aggregated$total_value, order.by = df_aggregated$date)

# View the result
xts_data

Quick Notes: #

• If your data has a monthly frequency instead of daily, adjust the frequency in the ts object to 12, and aggregate dates by month instead of day.
• Always check for missing or invalid dates with summary(df$date) before converting to a time series.

内容的提问来源于stack exchange，提问作者Saurabh