41.8 Customize your daily/automatic report

# Option A: One-liner automated HTML report (DataExplorer)
DataExplorer::create_report(tx, y = "median")  # sets target variable for focus

# Option B: Knit this Rmd on a schedule (cron/Task Scheduler)
# rmarkdown::render("eda_txhousing.Rmd", output_format = "html_document")
# Optionally explore additional EDA helpers
# install.packages(c("chronicle","descriptr"))
library(chronicle)
library(descriptr)
# Tip: Use these packages selectively; their APIs evolve.
# Prefer stable verbs above; consult vignettes for current functions.

41.8.1 Appendix: Small “gotchas” to keep in mind

  • For correlation heatmaps, ensure you’re not mixing units without scaling; use scale. = TRUE in PCA.
  • For discrete correlations, high-cardinality features (like many cities) can make matrices unwieldy. Thus, lump or select top-\(k\) first.
  • Imputation should be inside resampling to avoid leakage.
  • KPI checks (e.g., median vs volume/sales) are diagnostics, not proofs of error. Hence, investigate outliers before removal.