In akunuriYoshitha/BestTransform: Treats the Dataset for Best Possible Transformation

# df <- read.csv("//172.16.1.23/sli/Yoshitha/caterpillar-tube-pricing/data/competition_data/train_set.csv")
df <- read.csv("//172.16.1.23/sli/Yoshitha/HousePricing/missing_treated_77cols.csv", row.names=1)

Data Distribution

# data <- head(params$df, 100)
library(BestTransform)
# dist <- data_distribution(df, "cost")
dist <- data_distribution(params$df, params$dv)
dist[,5:8] <- round(dist[,5:8],2)
library(DT)

datatable(dist, rownames = F, filter = 'top', options = list(dom = "t", paging = FALSE)) %>% formatStyle(
  'is_dv',
  target = 'row',
  backgroundColor = styleEqual(TRUE, '#c2c2a3'),
  fontWeight = styleEqual(TRUE, "bold")
)

library(kableExtra)
# ks <- kable_styling(kable(dist), c("striped", "hover", "responsive"), full_width = F, fixed_thead = T, position = "center")
# scroll_box(ks, width = "1000px", height = "300px", fixed_thead = F)

Transformation Performance

# library(BestTransform)
# # trans <- BestTransform(df[1:500,], "cost")
# trans <- BestTransform(df[1:100,], "SalePrice")
# 
# ## Transformation Performance
# out <- trans$model_perf_metrics
# # out[,2:4] <- round(out[,2:4],2)
# best <- which(out[,"Rsquared"] == max(out[,"Rsquared"]))
# library(DT)
# datatable(out, rownames = F, options = list(dom = "t", paging = FALSE)) 
# %>% formatStyle(
#   'Rsquared',
#   target = 'row',
#   backgroundColor = styleEqual(best, '#00ff00'),
#   fontWeight = styleEqual(best, "bold")
# )

library(BestTransform)
# trans <- BestTransform(df[1:500,], "cost")
no_print <- function(x) 
{ sink(tempfile())
  on.exit(sink())
  invisible(force(x))
} 
trans <- no_print(BestTransform(params$df, params$dv))

## Transformation Performance
out <- trans$model_perf_metrics
ks <- kable_styling(kable(out), c("striped", "hover", "responsive"), full_width = T, fixed_thead = T, position = "left")

row_spec(ks, which(out[,"Rsquared"] == max(out[,"Rsquared"])), bold = T, color = "black", background = "#00ff00")

akunuriYoshitha/BestTransform documentation built on Feb. 19, 2021, 3:23 a.m.