This function add to the data frame generated by tidy_epma() columns outlier (boolean), pkint.L_est, pkint.H_est, mapint.L_est, beta, and mapint.H_est. Required columns in input are id, elint, phase, pk_t, beam, dwell, beam_map, mapint, mapint.L, mapint.H, pkint, pkint.L, pkint.H.

  phase = everything(),
  element = everything(),
  interval = c("prediction", "tukey"),
  method = c("rq", "lsfit", "median"),
  percentile = 0.99,
  fine_phase = NULL



A value returned by tidy_epma()

phase, element

Selected ones are referenced to detect outliers. Default selects everything. Tidy selection is available. For example c(Si, Ti) selects them, and c(-Si, -Ti) selects everything except them.


A type of the interval. Data points outside intervals are treated as outliers. If "prediction" (default), prediction intervals are used based on Poisson process. If `"tukey"``, conditional lower and upper whiskers are used where the lower is $Q_1/4 - 1.5 IQR$ and the upper is $Q_3/4 + 1.5 IQR$.


Applicable when interval = "prediction". If "rq" (default), quantile regression is performed (quantreg::rq()). If "lsfit", least square regression is performed (stats::lsfit()).


A percentile of predictive interaval. Applicable when interval = "prediction".


Deprecated as of qntmap > 0.4.0. Use phase instead.


When comparing data points of spot analysis and mapping analysis who share the same coordinates, it is expected that they analyze the same phases. However, this is unlikely when mapping analysis involves multi-phase pixels due to sizes of pixels and phases. This function removes such multi-phase pixels by finding outliers of mapping peak intensities.

See also