Merge pull request #25 from Nixtla/v2-version

chore: release v0.6.0

DESCRIPTION

@@ -25,10 +25,10 @@ Imports:
     ggplot2,
     httr2,
     lubridate,
+    purrr,
     rlang,
     tidyr,
-    tidyselect,
-    tsibble
+    tidyselect
 Suggests: 
     httptest2,
     knitr,

NAMESPACE

@@ -1,22 +1,13 @@
 # Generated by roxygen2: do not edit by hand
 
+export(.generate_output_dates)
 export(.get_api_key)
+export(.get_model_params)
 export(.level_from_quantiles)
 export(.make_request)
-export(.nixtla_client_cross_validation_distributed)
-export(.nixtla_client_cross_validation_seq)
-export(.nixtla_client_detect_anomalies_distributed)
-export(.nixtla_client_detect_anomalies_seq)
-export(.nixtla_client_forecast_distributed)
-export(.nixtla_client_forecast_seq)
-export(.nixtla_client_historic_distributed)
-export(.nixtla_client_historic_seq)
-export(.nixtla_data_prep)
-export(.partition_payload)
-export(.transform_output_dates)
+export(.r_frequency)
 export(.transient_errors)
 export(.validate_exogenous)
-export(date_conversion)
 export(infer_frequency)
 export(nixtla_client_cross_validation)
 export(nixtla_client_detect_anomalies)
@@ -25,8 +16,10 @@ export(nixtla_client_historic)
 export(nixtla_client_plot)
 export(nixtla_set_api_key)
 export(nixtla_validate_api_key)
+importFrom(dplyr,all_of)
 importFrom(dplyr,bind_rows)
 importFrom(dplyr,group_by)
+importFrom(dplyr,group_split)
 importFrom(dplyr,inner_join)
 importFrom(dplyr,mutate)
 importFrom(dplyr,n)
@@ -56,9 +49,9 @@ importFrom(httr2,request)
 importFrom(httr2,resp_status)
 importFrom(lubridate,ymd)
 importFrom(lubridate,ymd_hms)
+importFrom(purrr,map2_dfr)
 importFrom(rlang,.data)
 importFrom(tidyr,pivot_longer)
 importFrom(tidyselect,everything)
 importFrom(tidyselect,peek_vars)
 importFrom(tidyselect,starts_with)
-importFrom(tsibble,is_tsibble)

NEWS.md

@@ -1,3 +1,7 @@
+# nixtlar 0.6.0 
+
+- Development version. See release notes [here]()
+
 # nixtlar 0.5.4
 
 - Development version. See release notes [here](https://github.com/Nixtla/nixtlar/releases/tag/v0.5.4)

R/generate_output_dates.R

@@ -0,0 +1,49 @@
+#' Generate output dates for forecast method.
+#' This is a private function of 'nixtlar'
+#'
+#' @param df_info A data frame that is created by the forecast method with the last dates of every unique id.
+#' @param freq The frequency of the data, as a period or offset alias.
+#' @param h The forecast horizon.
+#'
+#' @return A data frame with dates for the forecast.
+#' @export
+#' @keywords internal
+#'
+#' @examples
+#' \dontrun{
+#'   dates_df <- .generate_output_dates(df_info, freq, h)
+#' }
+#'
+.generate_output_dates <- function(df_info, freq, h) {
+  new_dates <- lapply(1:nrow(df_info), function(i) {
+    start_date <- df_info$dates[i]
+    r_freq <- .r_frequency(freq)
+
+    if(freq == "QE") {
+      # End of quarter dates are: "YYY-03-31", "YYYY-06-30", "YYYY-09-30" and "YYYY-12-31".
+      dt <- seq(from = start_date, by = "quarter", length.out = h+1)
+      month <- lubridate::month(start_date)
+      if (month %in% c(3, 12)) {
+        dt <- ifelse(lubridate::month(dt) %in% c(7, 10), dt - lubridate::days(1), dt)
+      } else {
+        dt <- ifelse(lubridate::month(dt) %in% c(3, 12), dt + lubridate::days(1), dt)
+      }
+    } else if(freq == "ME") {
+      dt <- seq(from = start_date + lubridate::days(1), by = r_freq, length.out = h+1) - lubridate::days(1)
+    } else {
+      dt <- seq(from = start_date, by = r_freq, length.out = h+1)
+    }
+
+    dt[2:length(dt)]
+  })
+
+  dates_df <- data.frame(lapply(new_dates, as.POSIXct))
+
+  ids <- df_info$unique_id
+  if(inherits(df_info$unique_id, "numeric") | inherits(df_info$unique_id, "integer")){
+    ids <- as.character(df_info$unique_id)
+  }
+  names(dates_df) <- ids
+
+  return(dates_df)
+}

R/get_model_params.R

@@ -0,0 +1,46 @@
+#' Retrieve parameters for 'TimeGPT' model
+#' This is a private function of 'nixtlar'
+#'
+#' @param model Model to use, either "timegpt-1" or "timegpt-1-long-horizon".
+#' @param freq Frequency of the data.
+#'
+#' @return A list with the model's input size and horizon
+#' @export
+#' @keywords internal
+#'
+#' @examples
+#' \dontrun{
+#'   .get_model_params()
+#' }
+#'
+.get_model_params <- function(model, freq){
+
+  payload_params <- list(
+    model = model,
+    freq = freq
+  )
+
+  req <- httr2::request("https://api.nixtla.io/model_params") |>
+    httr2::req_headers(
+      "accept" = "application/json",
+      "content-type" = "application/json",
+      "authorization" = paste("Bearer", .get_api_key())
+    ) |>
+    httr2::req_user_agent("nixtlar") |>
+    httr2::req_body_json(data = payload_params) |>
+    httr2::req_retry(
+      max_tries = 6,
+      is_transient = .transient_errors
+    )
+
+  resp <- req |>
+    httr2::req_perform() |>
+    httr2::resp_body_json()
+
+  model_params <- list(
+    input_size = resp$detail$input_size,
+    horizon = resp$detail$horizon
+  )
+
+  return(model_params)
+}

R/infer_frequency.R

@@ -1,56 +1,70 @@
 #' Infer frequency of a data frame.
 #'
 #' @param df A data frame with time series data.
+#' @param freq The frequency of the data as specified by the user; NULL otherwise.
 #'
 #' @return The inferred frequency.
 #' @export
 #'
 #' @examples
 #' df <- nixtlar::electricity
-#' infer_frequency(df)
+#' freq <- NULL
+#' infer_frequency(df, freq)
 #'
-infer_frequency <- function(df){
-  freq <- NULL
-  dates <- sort(unique(df$ds))
-
-  # Check if it's hourly data
-  nchrs <- lapply(as.character(dates), nchar)
-  ntable <- sort(table(unlist(nchrs)))
-  nmode <- ntable[length(ntable)]
-  nmode <- as.numeric(names(nmode))
-
-  if(nmode > 10){
-    freq <- "H" # We'll assume hourly data
-    message("Frequency chosen: H")
+infer_frequency <- function(df, freq){
+  if(!is.null(freq)){
     return(freq)
   }
 
-  # If it's not hourly data, check the time differences in days
-  ddiff <- diff(as.Date(dates))
-  table <- sort(table(ddiff))
-  mode <- table[length(table)]
-  mode <- as.numeric(names(mode))
-
-  freq_list = list(
-    list(alias = "Y", value = c(365,366)),
-    list(alias = "Q", value = c(91,92)),
-    list(alias = "MS", value = c(30,31)),
-    list(alias = "W", value = c(7)),
-    list(alias = "D", value = c(1))
-  )
-
-  for(i in 1:length(freq_list)){
-    if(mode %in% freq_list[i][[1]]$value){
-      freq <- freq_list[i][[1]]$alias
+  num_chars <- nchar(as.character(df$ds[1]))
+
+  if(num_chars <= 10){
+    # assumes dates in format YYYY-MM-DD
+    dates <- lubridate::ymd(sort(unique(df$ds)))
+    dates_diff <- diff(dates)
+    dates_table <- table(dates_diff)
+    mode <- as.numeric(names(which.max(dates_table)))
+
+    freq_list = list(
+      list(alias = "Y", value = c(365,366)),
+      list(alias = "Q", value = c(91,92)),
+      list(alias = "MS", value = c(30,31)),
+      list(alias = "W", value = c(7)),
+      list(alias = "D", value = c(1))
+    )
+
+    for(item in freq_list){
+      if(mode %in% item$value){
+        freq <- item$alias
+        break
+      }
     }
+
+    message(paste0("Frequency chosen: ", freq))
+    return(freq)
+
+  }else{
+    # assumes dates in format YYYY-MM-DD hh:mm:ss
+    dates <- lubridate::ymd_hms(sort(unique(df$ds)))
+    dates_diff <- diff(dates)
+    dates_table <- table(dates_diff)
+    mode <- as.numeric(names(which.max(dates_table)))
+
+    units <- attr(dates_diff, "units")
+
+    freq <- switch(
+      units,
+      "hours" = ifelse(mode == 1, "h", paste0(mode, "h")),
+      "mins" = ifelse(mode == 1, "min", paste0(mode, "min")),
+      "secs" = ifelse(mode == 1, "s", paste0(mode, "s"))
+    )
+
+    message(paste0("Frequency chosen: ", freq))
+    return(freq)
   }
 
   if(is.null(freq)){
-    freq <- "D"
-    message("I'm not sure about the frequency of the data. Will default to daily (D). Please provide it if you know it.")
+    stop("I can't figure out the frequency of the data. Please specify it with the `freq` parameter")
   }
-
-  message(paste0("Frequency chosen: ", freq))
-
-  return(freq)
 }
+

R/nixtlaR-package.R

@@ -2,8 +2,10 @@
 "_PACKAGE"
 
 ## usethis namespace: start
+#' @importFrom dplyr all_of
 #' @importFrom dplyr bind_rows
 #' @importFrom dplyr group_by
+#' @importFrom dplyr group_split
 #' @importFrom dplyr inner_join
 #' @importFrom dplyr mutate
 #' @importFrom dplyr n
@@ -33,11 +35,11 @@
 #' @importFrom httr2 resp_status
 #' @importFrom lubridate ymd
 #' @importFrom lubridate ymd_hms
+#' @importFrom purrr map2_dfr
 #' @importFrom rlang .data
 #' @importFrom tidyr pivot_longer
 #' @importFrom tidyselect everything
 #' @importFrom tidyselect peek_vars
 #' @importFrom tidyselect starts_with
-#' @importFrom tsibble is_tsibble
 ## usethis namespace: end
 NULL