-
Notifications
You must be signed in to change notification settings - Fork 340
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #283 from facebookexperimental/lares
Clustering and functions split for exporting results
- Loading branch information
Showing
24 changed files
with
1,692 additions
and
1,043 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,204 @@ | ||
| # Copyright (c) Meta Platforms, Inc. and its affiliates. | ||
|
|
||
| # This source code is licensed under the MIT license found in the | ||
| # LICENSE file in the root directory of this source tree. | ||
|
|
||
| #################################################################### | ||
| #' Reduce number of models based on ROI clusters and minimum combined errors | ||
| #' | ||
| #' The \code{robyn_clusters()} function uses output from \code{robyn_run()}, | ||
| #' to reduce the amount of models and help the user pick up the best (lowest | ||
| #' combined error) of different kinds (clusters) of models. | ||
| #' | ||
| #' @inheritParams lares::clusterKmeans | ||
| #' @inheritParams hyper_names | ||
| #' @inheritParams robyn_outputs | ||
| #' @param input \code{robyn_export()}'s output or \code{pareto_aggregated.csv} results. | ||
| #' @param limit Integer. Top N results per cluster. If kept in "auto", will select k | ||
| #' as the cluster in which the WSS variance was less than 5\%. | ||
| #' @param weights Vector, size 3. How much should each error weight? | ||
| #' Order: nrmse, decomp.rssd, mape. The highest the value, the closer it will be scaled | ||
| #' to origin. Each value will be normalized so they all sum 1. | ||
| #' @param export Export plots into local files? | ||
| #' @param ... Additional parameters passed to \code{lares::clusterKmeans()}. | ||
| #' @author Bernardo Lares (bernardolares@@fb.com) | ||
| #' @examples | ||
| #' \dontrun{ | ||
| #' cls <- robyn_clusters(input = OutputCollect, | ||
| #' all_media = InputCollect$all_media, | ||
| #' k = 3, limit = 2, | ||
| #' weights = c(1, 1, 1.5)) | ||
| #' } | ||
| #' @export | ||
| robyn_clusters <- function(input, all_media = NULL, k = "auto", limit = 1, | ||
| weights = rep(1, 3), dim_red = "PCA", | ||
| quiet = FALSE, export = FALSE, | ||
| ...) { | ||
|
|
||
| if ("robyn_outputs" %in% class(input)) { | ||
| if (is.null(all_media)) { | ||
| aux <- colnames(input$mediaVecCollect) | ||
| all_media <- aux[-c(1, which(aux == "type"):length(aux))] | ||
| path <- input$plot_folder | ||
| } else path <- paste0(getwd(), "/") | ||
| # Pareto and ROI data | ||
| rois <- input$xDecompAgg | ||
| df <- .prepare_roi(rois, all_media = all_media) | ||
| } else { | ||
| if (all(c("solID", "mape", "nrmse", "decomp.rssd") %in% names(input)) & is.data.frame(input)) { | ||
| df <- .prepare_roi(input, all_media) | ||
| } else { | ||
| stop(paste( | ||
| "You must run robyn_export(..., clusters = TRUE) or", | ||
| "pass a valid data.frame (sames as pareto_aggregated.csv output)", | ||
| "in order to use robyn_clusters()" | ||
| )) | ||
| } | ||
| } | ||
|
|
||
| ignore <- c("solID", "mape", "decomp.rssd", "nrmse", "pareto") | ||
|
|
||
| # Auto K selected by less than 5% WSS variance (convergence) | ||
| min_clusters <- 3 | ||
| limit_clusters <- min(nrow(df) - 1, 30) | ||
| if ("auto" %in% k) { | ||
| cls <- tryCatch({ | ||
| clusterKmeans(df, k = NULL, limit = limit_clusters, ignore = ignore, dim_red = dim_red, quiet = TRUE, ...) | ||
| }, error = function(err) { | ||
| message(paste("Couldn't automatically create clusters:", err)) | ||
| return(NULL) | ||
| }) | ||
| #if (is.null(cls)) return(NULL) | ||
| min_var <- 0.05 | ||
| k <- cls$nclusters %>% | ||
| mutate(pareto = .data$wss/.data$wss[1], | ||
| dif = lag(.data$pareto) - .data$pareto) %>% | ||
| filter(.data$dif > min_var) %>% pull(.data$n) %>% max(.) | ||
| if (k < min_clusters) k <- min_clusters | ||
| if (!quiet) message(sprintf( | ||
| ">> Auto selected k = %s (clusters) based on minimum WSS variance of %s%%", | ||
| k, min_var*100)) | ||
| } | ||
|
|
||
| # Build clusters | ||
| stopifnot(k %in% min_clusters:30) | ||
| cls <- clusterKmeans(df, k, limit = limit_clusters, ignore = ignore, dim_red = dim_red, quiet = TRUE, ...) | ||
|
|
||
| # Select top models by minimum (weighted) distance to zero | ||
| top_sols <- .clusters_df(cls$df, weights) %>% | ||
| mutate(error = (.data$nrmse^2 + .data$decomp.rssd^2 + .data$mape^2)^-(1 / 2)) %>% | ||
| .crit_proc(limit) | ||
|
|
||
| output <- list( | ||
| # Data and parameters | ||
| data = mutate(cls$df, top_sol = .data$solID %in% top_sols$solID), | ||
| n_clusters = k, | ||
| errors_weights = weights, | ||
| # Within Groups Sum of Squares Plot | ||
| wss = cls$nclusters_plot, | ||
| # Grouped correlations per cluster | ||
| corrs = cls$correlations + labs(title = "ROI Top Correlations by Cluster", subtitle = NULL), | ||
| # Mean ROI per cluster | ||
| clusters_means = cls$means, | ||
| # Dim reduction clusters | ||
| clusters_PCA = cls[["PCA"]], | ||
| clusters_tSNE = cls[["tSNE"]], | ||
| # Top Clusters | ||
| models = top_sols, | ||
| plot_models_errors = .plot_topsols_errors(df, top_sols, limit, weights), | ||
| plot_models_rois = .plot_topsols_rois(top_sols, all_media, limit) | ||
| ) | ||
|
|
||
| if (export) { | ||
| fwrite(output$data, file = paste0(path, "pareto_clusters.csv")) | ||
| ggsave(paste0(path, "pareto_clusters_wss.png"), plot = output$wss, dpi = 500, width = 5, height = 4) | ||
| ggsave(paste0(path, "pareto_clusters_corr.png"), plot = output$corrs, dpi = 500, width = 7, height = 5) | ||
| db <- wrap_plots(output$plot_models_rois, output$plot_models_errors) | ||
| ggsave(paste0(path, "pareto_clusters_detail.png"), plot = db, dpi = 600, width = 12, height = 9) | ||
| } | ||
|
|
||
| return(output) | ||
|
|
||
| } | ||
|
|
||
|
|
||
| # ROIs data.frame for clustering (from xDecompAgg or pareto_aggregated.csv) | ||
| .prepare_roi <- function(x, all_media) { | ||
| check_opts(all_media, unique(x$rn)) | ||
| rois <- pivot_wider(x, id_cols = "solID", names_from = "rn", values_from = "roi_total") | ||
| rois <- removenacols(rois, all = FALSE) | ||
| rois <- select(rois, any_of(c("solID", all_media))) | ||
| errors <- distinct(x, .data$solID, .data$nrmse, .data$decomp.rssd, .data$mape) | ||
| rois <- left_join(rois, errors, "solID") %>% ungroup() | ||
| return(rois) | ||
| } | ||
|
|
||
| .min_max_norm <- function(x) (x - min(x)) / (max(x) - min(x)) | ||
|
|
||
| .clusters_df <- function(df, balance = rep(1, 3)) { | ||
| stopifnot(length(balance) == 3) | ||
| balance <- balance / sum(balance) | ||
| crit_df <- df %>% | ||
| # Force normalized values so they can be comparable | ||
| mutate( | ||
| nrmse = .min_max_norm(.data$nrmse), | ||
| decomp.rssd = .min_max_norm(.data$decomp.rssd), | ||
| mape = .min_max_norm(.data$mape) | ||
| ) %>% | ||
| # Balance to give more or less importance to each error | ||
| mutate( | ||
| nrmse = balance[1] / .data$nrmse, | ||
| decomp.rssd = balance[2] / .data$decomp.rssd, | ||
| mape = balance[3] / .data$mape | ||
| ) %>% | ||
| replace(., is.na(.), 0) %>% | ||
| group_by(.data$cluster) | ||
| return(crit_df) | ||
| } | ||
|
|
||
| .crit_proc <- function(df, limit) { | ||
| arrange(df, .data$cluster, desc(.data$error)) %>% | ||
| slice(1:limit) %>% | ||
| mutate(rank = row_number()) %>% | ||
| select(.data$cluster, .data$rank, everything()) | ||
| } | ||
|
|
||
| .plot_topsols_errors <- function(df, top_sols, limit = 1, balance = rep(1, 3)) { | ||
| balance <- balance / sum(balance) | ||
| left_join(df, select(top_sols, 1:3), "solID") %>% | ||
| mutate( | ||
| alpha = ifelse(is.na(.data$cluster), 0.5, 1), | ||
| label = ifelse(!is.na(.data$cluster), sprintf( | ||
| "[%s.%s]", .data$cluster, .data$rank | ||
| ), NA) | ||
| ) %>% | ||
| ggplot(aes(x = .data$nrmse, y = .data$decomp.rssd)) + | ||
| geom_point(aes(colour = .data$cluster, alpha = .data$alpha)) + | ||
| geom_text(aes(label = .data$label), na.rm = TRUE, hjust = -0.3) + | ||
| guides(alpha = "none", colour = "none") + | ||
| labs( | ||
| title = paste("Selecting Top", limit, "Performing Models by Cluster"), | ||
| subtitle = "Based on minimum (weighted) distance to origin", | ||
| x = "NRMSE", y = "DECOMP.RSSD", | ||
| caption = sprintf( | ||
| "Weights: NRMSE %s%%, DECOMP.RSSD %s%%, MAPE %s%%", | ||
| round(100 * balance[1]), round(100 * balance[2]), round(100 * balance[3]) | ||
| ) | ||
| ) + | ||
| theme_lares() | ||
| } | ||
|
|
||
| .plot_topsols_rois <- function(top_sols, all_media, limit = 1) { | ||
| top_sols %>% | ||
| mutate(label = sprintf("[%s.%s]\n%s", .data$cluster, .data$rank, .data$solID)) %>% | ||
| tidyr::gather("media", "roi", contains(all_media)) %>% | ||
| ggplot(aes(x = .data$media, y = .data$roi)) + | ||
| facet_grid(.data$label ~ .) + | ||
| geom_col() + | ||
| coord_flip() + | ||
| labs( | ||
| title = paste("ROIs on Top", limit, "Performing Models"), | ||
| x = NULL, y = "ROI per Media" | ||
| ) + | ||
| theme_lares() | ||
| } |
Oops, something went wrong.