Dynamic calculations with panel data in R-CodePudding

I have a panel data (individuals observed in different time periods) like df, with much more individuals in my dataset:

id <- c("1", "1", "1", "2", "2", "2")
t <- c("1", "2", "3", "1", "2", "3")
w <- c("0.17", "NA", "NA", "0.23", "NA", "NA")
alpha <- c("0.15", "0.15", "0.15", "0.15", "0.15", "0.15")
rho <- c("0.10", "0.21", "0.32", "0.12", "0.2", "0.08")

df <- data.frame(id, t, w, alpha, rho)

I would like to fill w following these mathematical dynamics: w_id_t = w_id_t-1 * alpha rho_id_t. However, I do not know how to move the outcome in such a way that appears in the following line, so I can continue with the dynamic calculations. The outcome should look like df_dynamics:

w_new <- c("0.17", "0.2355", "0.345", "0.23", "0.2345", "0.115")
df_dynamics <- data.frame(id, t, w_new, alpha, rho)

Any clue?

CodePudding user response：

The columns are all character class. We may need to convert to numeric first and then do a group by operation

library(dplyr)
type.convert(df, as.is = TRUE) %>%
    group_by(id) %>%
    mutate(w = coalesce(w, lag(first(w) * first(alpha)   lead(rho)))) %>% 
   ungroup
# A tibble: 6 × 5
     id     t     w alpha   rho
  <int> <int> <dbl> <dbl> <dbl>
1     1     1 0.17   0.15  0.1 
2     1     2 0.236  0.15  0.21
3     1     3 0.346  0.15  0.32
4     2     1 0.23   0.15  0.12
5     2     2 0.234  0.15  0.2 
6     2     3 0.114  0.15  0.08