Use the variables you create ```{r} weather_df |> group_by(name) |> mutate( temp_change = tmax - lag(tmax) ) |> summarize( sd_tmax_change = sd(temp_change, na.rm = TRUE), tmax_change_max = max(temp_change, na.rm = TRUE) ) ``` ```{r} weather_df |> group_by(name) |> mutate( temp_change = tmax - lag(tmax), change_rank = min_rank(desc(temp_change)) ) |> filter( change_rank < 2 ) ``` ```{r} weather_df |> group_by(name) |> mutate( temp_change = tmax - lag(tmax), ) |> filter( min_rank(desc(temp_change)) < 2 ) ``` ## Revisit PULSE ```{r} pulse_df = haven::read_sas("data/public_pulse_data.sas7bdat") |> janitor::clean_names() |> pivot_longer( bdi_score_bl:bdi_score_12m, names_to = "visit", names_prefix = "bdi_score_", values_to = "bdi" ) |> mutate(visit = fct_inorder(visit)) pulse_df |> ggplot(aes(x = visit, y = bdi)) + geom_boxplot() pulse_df |> group_by(visit) |> summarize( mean_bdi = mean(bdi, na.rm = TRUE), median_bdi = median(bdi, na.rm = TRUE) ) |> knitr::kable(digits = 2) ```