时间戳

首页 > 解决方案 > R rbind 在循环期间

r - R rbind 在循环期间

问题描述

此代码适用于所有内容......除了最后的 rbind 部分。我想从循环中获取结果并将它们组合成一个预测数据框。我试图创建一个空的临时数据框（我预测 12 个季度），但我得到一个“大小不等”的错误。

有人可以帮我吗？我快到了 ：-）

# Create actual data frame
Data_act <- subset(Data,Data$Type=="Actual")

# Create forecast frame with descriptive columns
Data_fc <- subset(Data[1:5],Data$Type=="Forecast")

# Create empty data frame for forecast results
Data_fc2 <- subset(Data[,6:9],Data$Type=="Forecast")
Data_fc2 <- Data_fc2[1:12,]
Data_fc3 <- Data_fc2 # temp table for loop

# Create list of unique forecast names
UniqueList <- unique(Data_act$forecast_name)

# Loop through unique list of forecast names
for(i in 1:length(UniqueList)){

    # Subset data for forecast name
    df <- subset(Data_act,Data_act$forecast_name==UniqueList[i])

    # Create time series objects
    dftsunits <- ts(df$Units,start = c(2015,1),frequency = 4)
    dftsasp <- ts(df$ASP,start = c(2015,1),frequency = 4)

    # Train forecasting models (Holt-Winters and ARIMA)
    FC_Units_HW <- hw(dftsunits,h=12)
    FC_Units_Arima <- auto.arima(dftsunits,lambda = 0) #lambda = 0 prevents     negative forecasts
    FC_ASP_HW <- hw(dftsasp,h=12)
    FC_ASP_Arima <- auto.arima(dftsasp,lambda = 0)

    # Forecast models for 12 quarters
    FC_Units_HW2 <- forecast(FC_Units_HW,h=12)
    FC_Units_Arima2 <- forecast(FC_Units_Arima,h=12)
    FC_ASP_HW2 <- forecast(FC_ASP_HW,h=12)
    FC_ASP_Arima2 <- forecast(FC_ASP_Arima,h=12)

    # Save results
    Data_fc3$Units_HW <- FC_Units_HW2$mean
    Data_fc3$Units_ARIMA <- FC_Units_Arima2$mean
    Data_fc3$ASP_HW <- FC_ASP_HW2$mean
    Data_fc3$ASP_ARIMA <- FC_ASP_Arima2$mean

    # Add results to master result data frame
    Data_fc2 <= rbind(Data_fc2,Data_fc3)
}

标签： rloopsrbind

解决方案

此代码有效。我很想知道如何提高效率。我需要在实际数据之后附加预测。

# Create actual data frame
Data_act <- subset(Data,Data$Type=="Actual")

# Create forecast frame with descriptive columns
Data_fc <- subset(Data[1:5],Data$Type=="Forecast")

# Create empty data frame for forecast results
Data_fc2 <- subset(Data[,6:9],Data$Type=="Forecast")
Data_fc2 <- Data_fc2[1:12,]
Data_fc3 <- Data_fc2 # temp table for loop

# Create list of unique forecast names
UniqueList <- unique(Data_act$forecast_name)

# Loop through unique list of forecast names and run time series
for(i in 1:length(UniqueList)){

    # Subset data for forecast name
    df <- subset(Data_act,Data_act$forecast_name==UniqueList[i])

    # Create time series objects
    dftsunits <- ts(df$Units,start = c(2015,1),frequency = 4)
    dftsasp <- ts(df$ASP,start = c(2015,1),frequency = 4)

    # Train forecasting models (Holt-Winters and ARIMA)
    FC_Units_HW <- hw(dftsunits,h=12)
    FC_Units_Arima <- auto.arima(dftsunits,lambda = 0) #lambda = 0 prevents negative forecasts
    FC_ASP_HW <- hw(dftsasp,h=12)
    FC_ASP_Arima <- auto.arima(dftsasp,lambda = 0)

    # Forecast models for 12 quarters
    FC_Units_HW2 <- forecast(FC_Units_HW,h=12)
    FC_Units_Arima2 <- forecast(FC_Units_Arima,h=12)
    FC_ASP_HW2 <- forecast(FC_ASP_HW,h=12)
    FC_ASP_Arima2 <- forecast(FC_ASP_Arima,h=12)

    # Save results
    Data_fc3$Units_HW <- FC_Units_HW2$mean
    Data_fc3$Units_ARIMA <- FC_Units_Arima2$mean
    Data_fc3$ASP_HW <- FC_ASP_HW2$mean
    Data_fc3$ASP_ARIMA <- FC_ASP_Arima2$mean

    # Add results to master result data frame
    Data_fc2 <- rbind(Data_fc2,Data_fc3)
}

Data_fc2 <- na.omit(Data_fc2)# Remove original 12 empty rows
Data_fc <- cbind(Data_fc,Data_fc2)
Data_Final <- rbind(Data_act,Data_fc)

推荐阅读