Select a subset of columns to compute stats for. See 'qsv select --help' for the format details. This is provided here because piping 'qsv select' into 'qsv stats' will disable the use of indexing.

Show all statistics available. This is provided here because piping 'qsv select' into 'qsv stats' will disable the use of indexing.

Infer data types only and do not compute statistics. Note that if you want to infer dates, you'll still need to use the --infer-dates and --dates-whitelist options.

Infer boolean data type. This automatically enables the --cardinality option. When a column's cardinality is 2, and the 2 values' first characters are 0/1, t/f & y/n case-insensitive, the data type is inferred as boolean.

Show the mode/s & antimode/s. Multimodal-aware. This requires loading all CSV data in memory.

Show the cardinality. This requires loading all CSV data in memory.

Show the median. This requires loading all CSV data in memory.

Shows the median absolute deviation (MAD). This requires loading all CSV data in memory.

Show the quartiles, the IQR, the lower/upper inner/outer fences and skewness. This requires loading all CSV data in memory.

Round statistics to <decimal_places>. Rounding is done following Midpoint Nearest Even (aka 'Bankers Rounding') rule. For dates - range, stddev & IQR are always at least 5 decimal places as they are reported in days, and 5 places gives us millisecond precision.

Include NULLs in the population size for computing mean and standard deviation.

Infer date/datetime datatypes. This is an expensive option and should only be used when you know there are date/datetime fields. Also, if timezone is not specified in the data, it'll be set to UTC.

The case-insensitive patterns to look for when shortlisting fields for date inferencing. i.e. if the field's name has any of these patterns, it is shortlisted for date inferencing. Set to 'all' to inspect ALL fields for date/datetime types. Ignored if --infer-dates is false. Note that false positive date matches WILL most likely occur when using 'all' as unix epoch timestamps are just numbers. Be sure to only use 'all' if you know the columns you're inspecting are dates, boolean or string fields. To avoid false positives, preprocess the file first with 'apply datefmt' to convert unix epoch timestamp columns to RFC3339 format.

Parse dates in dmy format. Otherwise, use mdy format. Ignored if --infer-dates is false.

Force recomputing stats even if valid precomputed stats cache exists.

The first row will be included in the count.

The field delimiter for reading CSV data. Must be a single character. (placeholder: ,)

Check if there is enough memory to load the entire CSV into memory using CONSERVATIVE heuristics. This option is ignored when computing default, streaming statistics, as it is not needed.