addressed check warnings and notes

.Rbuildignore

@@ -1,3 +1,7 @@
 ^.*\.Rproj$
 ^\.Rproj\.user$
-^\sample_data
+^sample_data$
+^index.html$
+^_config.yml$
+^README.md$
+^\.travis\.yml$

.travis.yml

@@ -0,0 +1,4 @@
+# R for travis: see documentation at https://docs.travis-ci.com/user/languages/r
+
+language: R
+cache: packages

DESCRIPTION

@@ -2,7 +2,7 @@ Package: FAIRforecast
 Type: Package
 Title: A Package for FAIR Forecast
 Version: 0.1.0
-Description: FAIR is a forecasting tool developed to support decision making in a retail environment. It provides multi-step-ahead sales forecasts at the category-store level, which are based on an interpretable and transparent model. These aspects make it an objective tool with which different promotional strategies (scenarios) can be compared and insights can be generated. Additionally, the FAIRforecast package provides plotting functions that generate figures showing the relative strength of the interactions between categories and important promotional variables. For more information, see vignette("FAIRforecast") and Gür Ali & Gürlek (2020).
+Description: FAIR is a forecasting tool developed to support decision making in a retail environment. It provides multi-step-ahead sales forecasts at the category-store level, which are based on an interpretable and transparent model. These aspects make it an objective tool with which different promotional strategies (scenarios) can be compared and insights can be generated. Additionally, the FAIRforecast package provides plotting functions that generate figures showing the relative strength of the interactions between categories and important promotional variables. For more information, see the vignette and original paper provided in the GitHub URL below.
 Authors@R: c(
     person("Özden", "Gür Ali", email = "[email protected]", role = "aut"),
     person("Ragip", "Gürlek", email = "[email protected]", role = c("aut", "cre")))
@@ -17,6 +17,8 @@ Imports:
     glmnet,
     plotly,
     plyr,
+	scales,
+	stats,
     tidyr
 RoxygenNote: 7.0.2
 Suggests: 

NAMESPACE

@@ -4,5 +4,4 @@ export(FAIR_predict)
 export(FAIR_train)
 export(cross_category)
 export(sample_data)
-export(var_cre)
 export(variable_importance)

R/fair_predict.R

@@ -84,7 +84,7 @@ FAIR_predict <- function(object, new_data, parallel = F) {
         return(rep(my_model, nrow(my_data)))
       if (is.null(my_model))
         return(rep(NA, nrow(my_data)))
-      return(suppressWarnings(predict.lm(my_model, my_data)))
+      return(suppressWarnings(stats::predict.lm(my_model, my_data)))
     }
 
     new_vars <- lapply(var_list, fit)

R/fair_train.R

@@ -110,7 +110,7 @@ FAIR_train <- function(train_data,
   for (t1 in t) {
     #block loop
     #deseasonalize data
-    is_val <- t1 != tail(t, 1)
+    is_val <- t1 != utils::tail(t, 1)
     if (is_val) {
       temp <- plyr::ddply(
         train_data,
@@ -201,15 +201,15 @@ FAIR_train <- function(train_data,
   best_pars <- function(my_data) {
     my_data$alpha <- as.numeric(as.character(my_data$alpha))
     my_data$lambda <- as.numeric(as.character(my_data$lambda))
-    parameter_ev1 <- aggregate(
+    parameter_ev1 <- stats::aggregate(
       my_data$RMSE,
       by = list(my_data$alpha, my_data$lambda),
       FUN = "mean",
       na.rm = TRUE,
       drop = TRUE
     )
     names(parameter_ev1)[ncol(parameter_ev1)] <- "mean"
-    parameter_ev2 <- aggregate(
+    parameter_ev2 <- stats::aggregate(
       my_data$RMSE,
       by = list(my_data$alpha, my_data$lambda),
       FUN = "sd",

R/others.R

@@ -98,21 +98,21 @@ deseason <- function(my_data,
   var_list <- c(sales, marketing)
   fit <- function(my_var) {
     train <- my_data[my_data[, time_id] < t1, c(my_var, seasonality)]
-    if (sd(train[, my_var]) == 0){
+    if (stats::sd(train[, my_var]) == 0){
       s1_models[[my_var]] <<- train[1, my_var]
       return(my_data[, my_var])
       #If there is no variance in the training data, return
       #training + validation as it is
     } else{
-      myformula <- as.formula(paste0("log(", my_var, "+1) ~ ."))
-      my_model <- lm(formula = myformula,
+      myformula <- stats::as.formula(paste0("log(", my_var, "+1) ~ ."))
+      my_model <- stats::lm(formula = myformula,
                      data = train,
                      model = F)
       s1_models[[my_var]] <<- my_model
       #above line ensures obs until t1+horizon-1 are predicted.
       #If the data is not validation (training), it is same as t1-1 since
       #all points are less than t1
-      return(log(my_data[, my_var] + 1) - predict.lm(my_model, my_data))
+      return(log(my_data[, my_var] + 1) - stats::predict.lm(my_model, my_data))
     }
   }
   new_vars <- lapply(var_list, fit)
@@ -123,7 +123,7 @@ deseason <- function(my_data,
   my_list <- list(my_data, s1_models)
   return(my_list)
 }
-#'@export
+
 var_cre <- function(my_data,
                     category,
                     category_list,
@@ -191,7 +191,7 @@ step2 <-
     }
 
     train <- my_data[my_data[, time_id] < t1,]
-    train <- train[complete.cases(train[, marketing]),]
+    train <- train[stats::complete.cases(train[, marketing]),]
 
     if (is_val) {
       test <- my_data[!my_data[, time_id] < t1,]
@@ -259,7 +259,7 @@ step2 <-
 
 time_series <- function(my_data, store, time_id, category) {
   sum_func <- function(x) if (all(is.na(x))) NA_integer_  else sum(x)
-  series <- aggregate(
+  series <- stats::aggregate(
     my_data$residual_2,
     by = list(my_data[, store], my_data[, time_id], my_data[, category]),
     FUN = sum_func,
@@ -292,7 +292,7 @@ time_series <- function(my_data, store, time_id, category) {
 
 step3 <- function(x, time_id, frequency) {
   x <- x[order(x[, time_id]),]
-  my_series <- ts(x$residual_2, frequency = frequency)
+  my_series <- stats::ts(x$residual_2, frequency = frequency)
   my_model <- forecast::stlm(forecast::na.interp(my_series))
   x$STL <- as.numeric(my_model$fitted)
   return(list(x, my_model))
@@ -371,7 +371,7 @@ cross_category <- function(object){
     my_mat <- my_mat[my_mat[, 'variable'] %in% cc_list, ]
     my_mat$category <- substr(my_mat$variable, 1,
                               regexpr("_", my_mat$variable) - 1)
-    my_mat <- aggregate(my_mat$magnitude, list(my_mat$category), FUN = sum)
+    my_mat <- stats::aggregate(my_mat$magnitude, list(my_mat$category), FUN = sum)
     colnames(my_mat) <- c('Category', 'magnitude')
     for(j in 1:nrow(my_mat)){
       cross_mat[names(models)[i], my_mat[j, 'Category']] <-
@@ -383,8 +383,7 @@ cross_category <- function(object){
   p <- plotly::layout(
     plotly::plot_ly(x = colnames(cross_mat), y = rownames(cross_mat),
                     z = cross_mat,
-                    colorscale = scales::col_numeric("Blues", domain = NULL)(unique(scales::rescale(c(volcano)))),
-                    type = "heatmap"),
+colors = "Reds",                    type = "heatmap"),
     title = "Cross-category effects",
     xaxis = list(title = "Influencial Category"),
     yaxis = list(title = "Influenced Category")

vignettes/FAIRforecast.Rmd

@@ -128,29 +128,6 @@ cross_category(my_model)
 ```
 
 
-```{r eval=FALSE, include=FALSE}
-### This part is not included in the vignette. Because, the current version of FAIR does not have a variable to distinguish the scenarios with the same store-category-week combination. Also, with the sample data we have, the other categories also should be provided to the predict function because of cc_marketing. Here is the paragraph I wrote for this example:
-### The user can input a set of scenarios to `FAIR_predict` function to guide a decision. For example, the below code compares the sales for a base scenario, an agressive-ad scenario, and a scenario where the same agressive ad strategy is followed on a holiday.
-
-# The base scenario. We take a random observation from the test data.
-scenarios <- test_data[3757, ]
-scenarios
-
-# The agressive-ad scenario. We change the average ad in the category.
-scenarios[2, ] <- scenarios[1, ]
-scenarios[2, "ad"] <- 0.30
-
-# Agressive-ad-on-holiday scenario. We take an observation on a holiday for the same store-category and change its marketing variables to the those of scenario 2.
-scenarios[3, ] <- test_data[3766, ]
-marketing_variables = c("price", "display", "discount", "dist_5", "dist_10",
-                        "dist_15", "dist_20", "dist_30", "dist_40", "dist_50",
-                        "dist_100", "ad")
-scenarios[3, marketing_variables] <- scenarios[2, marketing_variables]
-
-FAIR_predict(my_model, scenarios)
-```
-
-
 ## References
 
 Gür Ali, Ö. and Gürlek, R. (2019) Automatic Interpretable Retail Forecasting with Promotional Scenarios