如何通过沿用前值增加xts对象的观测数量？

Hey there! I get exactly what you're dealing with—cleaning minute-level price data that only logs values when trades happen, and needing to fill in the quiet minutes with the last recorded price. Those Stack Overflow posts about reducing xts periodicity don't apply here because we're doing the opposite: adding missing time points and carrying forward the last valid value.

Here's a straightforward, R-based solution using xts tools that fits your use case perfectly:

Step-by-Step Implementation #

• First, we need to create a complete minute-by-minute time sequence that spans the entire range of your original data. This gives us a template with every minute accounted for, even the ones with no trades.
• Merge your original trade data with this full time sequence—this will introduce NA values for all minutes where no trade occurred.
• Use the na.locf() function (short for "Last Observation Carried Forward") to fill those NAs with the most recent valid price.

Example Code #

Assuming your original data is an xts object named trade_prices (with timestamps for trade minutes and a price column):

# Generate a full minute-level time sequence covering your data's range
full_time_seq <- seq(start(trade_prices), end(trade_prices), by = "min")
full_xts_template <- xts(, order.by = full_time_seq)

# Merge original data with the full template to create gaps (NAs)
merged_data <- merge(trade_prices, full_xts_template)

# Fill NAs with the last observed price
filled_price_data <- na.locf(merged_data)

Quick Notes #

• na.locf() works exactly as you need it to: it propagates the last non-NA value forward through subsequent rows, which matches your requirement that "no trade = price stays the same as the last minute".
• If you have any leading NAs (unlikely in your case since trades start at some point), you can add the na.rm = FALSE argument to keep them, but in most trade data scenarios, you won't need this.

内容的提问来源于stack exchange，提问作者stollenm