Error in UseMethod("model") - forecasting error using time series

Hello,

I have a rds file that has data that looks like this (I named it rds_intake):

I have data for all 3 providers from January 1, 2021 to March 23, 2023. So I have almost 2 whole fiscal years worth of data (2021-2022 and 2022-2023). Here, fiscal year and quarter is considered:

Jan 01, 2021 - March 31, 2021 = 2020-2021 (fiscal_yr); Q4 (quarter)
Apr 01, 2021 - June 30, 2021 = 2021-2022; Q1
Jul 01, 2021 - Sep 30, 2021 = 2021-2022; Q2
Oct 01, 2021 - Dec 31, 2021 = 2021-2022; Q3
Jan 01, 2022 - March 31, 2022 = 2021-2022; Q4
Apr 01, 2022 - June 30, 2022 = 2022-2023; Q1...and so forth

I used the following function to aggregate the client_id because I would like to forecast the number of clients for the next 3 years so I used the following function to create this:

p_intake <- rds_intake %>% mutate(start_month = yearmonth(start_month)) %>% group_by(provider, start_month) %>% summarize(intake = n_distinct(client_id))

then I plotted the data that looks like this:

Then I tried the forecast models because I want to forecast what the next 3 years would look like so I used this function:

library(forecast)
library(thief)
library(fable)

intake_fit <- p_intake %>% model( seasonal_naive = SNAIVE(intake), naive_p = NAIVE(intake), drift_p = RW(intake ~ drift()), mean_p = mean(intake) )

intake_fit <- p_intake %>% model(hw = ETS(intake~ season("A"))) %>% tidy()

however, this is the error I get:

Error in UseMethod("model") :
no applicable method for 'model' applied to an object of class "c('grouped_df', 'tbl_df', 'tbl', 'data.frame')"

If someone could please help me fix this issue, I would really appreciate your help. Thank you

The data, not just a screenshot, is needed to complete a reprex. See the FAQ.

Also consider using {fpp3} and its companion text with the rest of the tools of the same generation as the {fable} package, particularly tsibble() which is needed for the modeling functions you are using. These are by the same team who produced {forecast} and they intend it to replace that generation of time series tools.