Merge pull request #936 from facebookexperimental/bl_branch

fix: mediaVecCollect's organic vars weren't being multiplied by coef values + InputCollect print when no prophet country set

Author: laresbernardo

R/DESCRIPTION

@@ -1,7 +1,7 @@
 Package: Robyn
 Type: Package
 Title: Semi-Automated Marketing Mix Modeling (MMM) from Meta Marketing Science 
-Version: 3.10.6.9001
+Version: 3.10.6.9002
 Authors@R: c(
     person("Gufeng", "Zhou", , "gufeng@meta.com", c("cre","aut")),
     person("Bernardo", "Lares", , "laresbernardo@gmail.com", c("aut")),

R/R/checks.R

@@ -243,7 +243,7 @@ check_paidmedia <- function(dt_input, paid_media_vars, paid_media_signs, paid_me
   check_vector(paid_media_vars)
   check_vector(paid_media_signs)
   check_vector(paid_media_spends)
-  mediaVarCount <- length(paid_media_vars)
+  expVarCount <- length(paid_media_vars)
   spendVarCount <- length(paid_media_spends)
 
   temp <- paid_media_vars %in% names(dt_input)
@@ -261,7 +261,7 @@ check_paidmedia <- function(dt_input, paid_media_vars, paid_media_signs, paid_me
     ))
   }
   if (is.null(paid_media_signs)) {
-    paid_media_signs <- rep("positive", mediaVarCount)
+    paid_media_signs <- rep("positive", expVarCount)
   }
   if (!all(paid_media_signs %in% OPTS_PDN)) {
     stop("Allowed values for 'paid_media_signs' are: ", paste(OPTS_PDN, collapse = ", "))
@@ -272,7 +272,7 @@ check_paidmedia <- function(dt_input, paid_media_vars, paid_media_signs, paid_me
   if (length(paid_media_signs) != length(paid_media_vars)) {
     stop("Input 'paid_media_signs' must have same length as 'paid_media_vars'")
   }
-  if (spendVarCount != mediaVarCount) {
+  if (spendVarCount != expVarCount) {
     stop("Input 'paid_media_spends' must have same length as 'paid_media_vars'")
   }
   is_num <- unlist(lapply(dt_input[, paid_media_vars], is.numeric))
@@ -295,7 +295,7 @@ check_paidmedia <- function(dt_input, paid_media_vars, paid_media_signs, paid_me
   }
   return(invisible(list(
     paid_media_signs = paid_media_signs,
-    mediaVarCount = mediaVarCount,
+    expVarCount = expVarCount,
     paid_media_vars = paid_media_vars
   )))
 }
@@ -706,7 +706,7 @@ check_InputCollect <- function(list) {
   names_list <- c(
     "dt_input", "paid_media_vars", "paid_media_spends", "context_vars",
     "organic_vars", "all_ind_vars", "date_var", "dep_var",
-    "rollingWindowStartWhich", "rollingWindowEndWhich", "mediaVarCount",
+    "rollingWindowStartWhich", "rollingWindowEndWhich",
     "factor_vars", "prophet_vars", "prophet_signs", "prophet_country",
     "intervalType", "dt_holidays"
   )

R/R/clusters.R

@@ -41,7 +41,7 @@
 robyn_clusters <- function(input, dep_var_type,
                            cluster_by = "hyperparameters",
                            all_media = NULL,
-                           k = "auto", wss_var = 0.05, limit = 1,
+                           k = "auto", wss_var = 0.06, limit = 1,
                            weights = rep(1, 3), dim_red = "PCA",
                            quiet = FALSE, export = FALSE, seed = 123,
                            ...) {
@@ -315,8 +315,7 @@ errors_scores <- function(df, balance = rep(1, 3), ts_validation = TRUE, ...) {
         select(any_of(c("solID", all_media)))
     }
     errors <- distinct(
-      x, .data$solID, .data$nrmse, .data$nrmse_test,
-      .data$nrmse_train, .data$decomp.rssd, .data$mape
+      x, .data$solID, starts_with("nrmse"), .data$decomp.rssd, .data$mape
     )
     outcome <- left_join(outcome, errors, "solID") %>% ungroup()
   } else {

R/R/inputs.R

@@ -231,11 +231,10 @@ robyn_inputs <- function(dt_input = NULL,
     context <- check_context(dt_input, context_vars, context_signs)
     context_signs <- context$context_signs
 
-    ## Check paid media variables (set mediaVarCount and maybe transform paid_media_signs)
+    ## Check paid media variables (and maybe transform paid_media_signs)
     if (is.null(paid_media_vars)) paid_media_vars <- paid_media_spends
     paidmedia <- check_paidmedia(dt_input, paid_media_vars, paid_media_signs, paid_media_spends)
     paid_media_signs <- paidmedia$paid_media_signs
-    mediaVarCount <- paidmedia$mediaVarCount
     exposure_vars <- paid_media_vars[!(paid_media_vars == paid_media_spends)]
 
     ## Check organic media variables (and maybe transform organic_signs)
@@ -309,7 +308,6 @@ robyn_inputs <- function(dt_input = NULL,
       paid_media_signs = paid_media_signs,
       paid_media_spends = paid_media_spends,
       paid_media_total = paid_media_total,
-      mediaVarCount = mediaVarCount,
       exposure_vars = exposure_vars,
       organic_vars = organic_vars,
       organic_signs = organic_signs,
@@ -435,7 +433,8 @@ Adstock: {x$adstock}
     windows = paste(x$window_start, x$window_end, sep = ":"),
     custom_params = if (length(x$custom_params) > 0) paste("\n", flatten_hyps(x$custom_params)) else "None",
     prophet = if (length(x$prophet_vars) > 0) {
-      sprintf("%s on %s", paste(x$prophet_vars, collapse = ", "), x$prophet_country)
+      sprintf("%s on %s", paste(x$prophet_vars, collapse = ", "),
+              ifelse(!is.null(x$prophet_country), x$prophet_country, "data"))
     } else {
       "\033[0;31mDeactivated\033[0m"
     },
@@ -622,7 +621,7 @@ robyn_engineering <- function(x, quiet = FALSE, ...) {
     mediaCostFactor <- colSums(subset(dt_inputRollWind, select = paid_media_spends), na.rm = TRUE) /
       colSums(subset(dt_inputRollWind, select = paid_media_vars), na.rm = TRUE)
 
-    for (i in 1:InputCollect$mediaVarCount) {
+    for (i in seq_along(paid_media_spends)) {
       if (exposure_selector[i]) {
         # Run models (NLS and/or LM)
         dt_spendModInput <- subset(dt_inputRollWind, select = c(paid_media_spends[i], paid_media_vars[i]))

R/R/pareto.R

@@ -407,7 +407,7 @@ robyn_pareto <- function(InputCollect, OutputModels,
         )
       }
       dt_transformSaturationDecomp <- dt_transformSaturation
-      for (i in 1:InputCollect$mediaVarCount) {
+      for (i in seq_along(InputCollect$all_media)) {
         coef <- plotWaterfallLoop$coef[plotWaterfallLoop$rn == InputCollect$all_media[i]]
         dt_transformSaturationDecomp[InputCollect$all_media[i]] <- coef *
           dt_transformSaturationDecomp[InputCollect$all_media[i]]
@@ -418,7 +418,7 @@ robyn_pareto <- function(InputCollect, OutputModels,
 
       ## Reverse MM fitting
       # dt_transformSaturationSpendReverse <- copy(dt_transformAdstock[, c("ds", InputCollect$all_media), with = FALSE])
-      # for (i in 1:InputCollect$mediaVarCount) {
+      # for (i in seq_along(InputCollect$paid_media_spends)) {
       #   chn <- InputCollect$paid_media_vars[i]
       #   if (chn %in% InputCollect$paid_media_vars[InputCollect$exposure_selector]) {
       #     # Get Michaelis Menten nls fitting param