是什么导致R中的预测函数出现此错误？

Question

我收到一条错误消息： “错误：变量'date'的类型为“ other”，但类型为“ numeric”的 提供”

我尝试预测健身模型的预计日期，但是似乎 可能与我的数据集有关，但我不知道在哪里。

问题2预测3-30到4-8的阳性病例和预测的转化间隔**需要与原始量表进行比较

pred_data = data[data$date>="2020-03-30" & data$date<="2020-04-08",]
pred_data$pos_trans = (pred_data$positive^lam-1)/lam
pred_data$pred = predict(fit_trans,newdata = pred_data)
predict(fit_trans, newdata = pred_data,interval = "predict")
pred_data


data_Test$pos_trans_pred = predict(fit_trans, newdata = data_Test)
pred_error = data_Test$pos_trans_pred-data_Test$pos_trans
pred_error_trans = mean(pred_error^2)
pred_error_trans
pred_error
plot(US$date, US$pos_trans,ylim = c(25,125))
lines(data_Train$date,fit_trans$fitted.values,col=2)
lines(data_Test$date, data_Test$pos_trans_pred, col=3)

1. 错误：变量'date'符合类型“ other”，但提供了“ numeric”类型

2. 我的原始开始日期如下：

##Model w/3/16-29
##Train 3/16-3/29
##Test 3/30-4/8

    library(data.table)
    data = fread('https://covidtracking.com/api/us/daily.csv',data.table=FALSE)
    days = 24
    loc = which(data[,1]=='20200408')
    US = data[loc:(loc+days-1),c(1,3)]
    US$date = as.Date(as.character(US$date), format = '%Y%m%d')
    fit = lm(positive ~ date, data = US)
    train_Ind = 11:24
    data_Train = US[train_Ind,]
    data_Test = US[-train_Ind,]
    fit = lm(positive ~ date, data = data_Train)
    library(MASS)
    boxcox_fit = boxcox(fit)
    lam = boxcox_fit$x[which.max(boxcox_fit$y)]
    data_Train$pos_trans = (data_Train$positive^lam-1)/lam
    data_Test$pos_trans = (data_Test$positive^lam-1)/lam
    US$pos_trans = (US$positive^lam-1)/lam
    fit_trans = lm(pos_trans ~ date, data = data_Train)