R语言每日数据autoplot轴标签与ggseasonplot报错问题求助

Hey there! Let's tackle your two R time series issues one by one.

1. Fixing X-axis to show dates instead of row numbers in autoplot #

The root problem here is that converting your xts object to a ts object with as.ts() strips away the actual date information—replacing it with a sequential row index (start=1, end=742). The autoplot function for ts objects uses this index by default for the X-axis.

Here are two simple fixes:

Option 1: Use your xts object directly #

The xts package plays nicely with ggfortify's autoplot, so you can skip the ts conversion entirely:

library(ggfortify)
autoplot(cw_x[,2:5], facets = TRUE) + 
  ylab("parameters") +
  xlab("Date")

This will automatically pull the date index from your xts object for the X-axis.

Option 2: Plot directly from the original data frame with ggplot2 #

If you prefer working with the raw data frame, reshape it to long format and use ggplot2 for full control over axes:

library(ggplot2)
library(tidyr)

# Reshape data to long format for faceted plotting
cw_long <- cw %>%
  select(date, amp, brg_de_tmp, brg_nde_tmp, kg) %>%
  pivot_longer(cols = -date, names_to = "Parameter", values_to = "Value")

ggplot(cw_long, aes(x = date, y = Value)) +
  geom_line() +
  facet_wrap(~Parameter, scales = "free_y") +
  ylab("parameters") +
  xlab("Date")

2. Fixing "Data are not seasonal" error for ggseasonplot #

The ggseasonplot function (from the forecast package) needs a time series with a defined seasonal frequency to detect patterns. Your current cw_ts has Frequency = 1, which tells R it's annual data with no seasonality—hence the error.

For daily data, we need to set the correct frequency or use a more date-friendly time series structure. Here are two approaches:

Option 1: Create a proper ts object with daily frequency #

First, define the start date as a year-month-day tuple, then set frequency = 365 (for non-leap years):

library(forecast)

# Get start date components
start_date <- as.Date("2016-04-01")
start <- c(year(start_date), month(start_date), day(start_date))

# Create ts object with daily frequency
cw_daily_ts <- ts(cw[,2:13], start = start, frequency = 365)

# Plot seasonplot for a single parameter (e.g., amp)
ggseasonplot(cw_daily_ts[,1], year.labels=TRUE, year.labels.left=TRUE) + 
  ylab("amp") + 
  ggtitle("Seasonal plot: Pump Amplitude")

# Polar seasonal plot
ggseasonplot(cw_daily_ts[,1], polar=TRUE) + 
  ylab("amp") + 
  ggtitle("Polar seasonal plot: Pump Amplitude")

Note: ggseasonplot works best with single time series. To plot multiple parameters, you'll need to loop through them or reshape your data.

Option 2: Use tsibble for intuitive daily time series handling #

The tsibble package is built for tidy time series, and pairs with feasts (which has a flexible gg_season function):

library(tsibble)
library(feasts)

# Convert data frame to tsibble (date as the index)
cw_tsibble <- as_tsibble(cw, index = date)

# Seasonal plot for a single parameter
cw_tsibble %>%
  gg_season(amp, year.labels = TRUE) +
  ylab("amp") +
  ggtitle("Seasonal plot: Pump Amplitude")

# Polar seasonal plot
cw_tsibble %>%
  gg_season(amp, polar = TRUE) +
  ylab("amp") +
  ggtitle("Polar seasonal plot: Pump Amplitude")

# Plot multiple parameters with facets
cw_tsibble_long <- cw_tsibble %>%
  select(date, amp, brg_de_tmp, brg_nde_tmp, kg) %>%
  pivot_longer(cols = -date, names_to = "Parameter", values_to = "Value")

cw_tsibble_long %>%
  gg_season(Value, year.labels = TRUE) +
  facet_wrap(~Parameter, scales = "free_y") +
  ylab("parameters") +
  ggtitle("Seasonal plots: Pump Parameters")

This approach avoids frequency-setting headaches and is more flexible for daily data.

内容的提问来源于stack exchange，提问作者Avijit Mallick