dataframe-toolkit 0.5.1__tar.gz → 0.7.0__tar.gz

{dataframe_toolkit-0.5.1 → dataframe_toolkit-0.7.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: dataframe-toolkit
-Version: 0.5.1
+Version: 0.7.0
 Summary: A command-line toolkit for DataFrame analysis and manipulation
 Keywords: dataframe,statistics,data analysis,CLI,TSV,pandas
 Author: Gunnar Pálsson

dataframe_toolkit-0.7.0/pyproject.toml

@@ -0,0 +1,82 @@
+[project]
+name = "dataframe-toolkit"
+version = "0.7.0"
+description = "A command-line toolkit for DataFrame analysis and manipulation"
+readme = "README.md"
+keywords = [
+    "dataframe",
+    "statistics",
+    "data analysis",
+    "CLI",
+    "TSV",
+    "pandas",
+]
+classifiers = [
+    "Development Status :: 3 - Alpha",
+    "Intended Audience :: Science/Research",
+    "Intended Audience :: Developers",
+    "Topic :: Scientific/Engineering :: Information Analysis",
+    "Topic :: Utilities",
+    "Programming Language :: Python :: 3",
+]
+requires-python = ">=3.12"
+dependencies = [
+    "pandas>=2.1",
+    "numpy>=1.24",
+    "pyarrow>=12.0",
+    "scipy>=1.10",
+    "statsmodels>=0.14",
+    "duckdb>=0.8",
+    "matplotlib>=3.7",
+    "seaborn>=0.12",
+    "tqdm>=4.65",
+]
+
+[[project.authors]]
+name = "Gunnar Pálsson"
+email = "hemingur@gmail.com"
+
+[project.license]
+text = "MIT"
+
+[project.urls]
+Homepage = "https://github.com/hemingur/dataframe-toolkit"
+Repository = "https://github.com/hemingur/dataframe-toolkit"
+
+[project.scripts]
+dftk = "dftk.cli:main"
+
+[build-system]
+requires = ["uv_build>=0.10.4,<0.11.0"]
+build-backend = "uv_build"
+
+[tool.uv.build-backend]
+module-name = "dftk"
+
+[tool.pytest.ini_options]
+testpaths = ["tests"]
+
+[tool.ruff]
+line-length = 88
+target-version = "py312"
+
+[tool.ruff.lint]
+select = [
+    "E",
+    "W",
+    "F",
+    "I",
+    "UP",
+    "B",
+]
+
+[tool.ruff.lint.isort]
+known-first-party = ["dftk"]
+
+[dependency-groups]
+dev = [
+    "pre-commit>=4.6.0",
+    "pytest>=8.0",
+    "pytest-cov>=7.1.0",
+    "ruff>=0.15.12",
+]

dataframe_toolkit-0.5.1/pyproject.toml → dataframe_toolkit-0.7.0/pyproject.toml.orig

@@ -1,6 +1,6 @@
 [project]
 name = "dataframe-toolkit"
-version = "0.5.1"
+version = "0.7.0"
 description = "A command-line toolkit for DataFrame analysis and manipulation"
 readme = "README.md"
 authors = [

{dataframe_toolkit-0.5.1 → dataframe_toolkit-0.7.0}/src/dftk/cli.py

@@ -8,6 +8,8 @@ logic and never needs to change when a new subcommand is added.
 
 import argparse
 import logging
+import os
+import signal
 import sys
 
 from dftk import __version__
@@ -47,6 +49,13 @@ def main() -> None:
 
     try:
         args.command_instance.execute(args)
+        sys.stdout.flush()
+    except BrokenPipeError:
+        # The reader closed the pipe early (e.g. `dftk ... | head`).  Exit
+        # quietly with the SIGPIPE status like cat/grep, and point stdout at
+        # devnull so the interpreter's final flush doesn't raise again.
+        os.dup2(os.open(os.devnull, os.O_WRONLY), sys.stdout.fileno())
+        sys.exit(128 + signal.SIGPIPE)
     except FileNotFoundError as exc:
         print(f"File error: {exc}", file=sys.stderr)
         sys.exit(1)

{dataframe_toolkit-0.5.1 → dataframe_toolkit-0.7.0}/src/dftk/commands/__init__.py

@@ -8,6 +8,7 @@
 
 from dftk.commands.annotate_cmd import AnnotateCommand
 from dftk.commands.binx_cmd import BinxCommand
+from dftk.commands.chi_cmd import ChiCommand
 from dftk.commands.clean_cmd import CleanCommand
 from dftk.commands.concat_cmd import ConcatCommand
 from dftk.commands.corr_cmd import CorrCommand
@@ -80,6 +81,7 @@ COMMAND_GROUPS: list[tuple[str, list]] = [
             ScatCommand(),
             LineCommand(),
             HistCommand(),
+            ChiCommand(),
         ],
     ),
     (

{dataframe_toolkit-0.5.1 → dataframe_toolkit-0.7.0}/src/dftk/commands/annotate_cmd.py

@@ -25,6 +25,10 @@ Usage examples:
   dftk eval data.tsv -f "z = x + y" -o result.parquet \\
       --meta genome=hg38 --meta source=my_pipeline
 
+  # Same, built from a bundled dataset:
+  dftk dataset iris -o | dftk eval ... -f "z = sepal_length + petal_length" \\
+      -o result.parquet --meta genome=hg38 --meta source=my_pipeline
+
 Annotations propagate through the parquet pipe: a file annotated at
 creation is re-annotated on every subsequent -o write unless explicitly
 overridden.

{dataframe_toolkit-0.5.1 → dataframe_toolkit-0.7.0}/src/dftk/commands/binx_cmd.py

@@ -74,8 +74,10 @@ USEVALUE (--usevalue)
 EXAMPLES
 --------
   dftk binx data.tsv -c age -b 0:100:10
-  dftk binx data.tsv -c score -b 0,25,50,75,100 -d quartile --usevalue m
-  dftk binx data.tsv -c x -b -3:4:1 -d x_bin -o | dftk stat -g x_bin -s y
+  dftk dataset tips -o | dftk binx ... -c total_bill -b 0:60:10
+  dftk dataset tips -o | dftk binx ... -c tip -b 0,2,4,6,10 -d tip_bracket --usevalue m
+  dftk dataset tips -o | dftk binx ... -c total_bill -b 0:60:10 -d bill_bin -o \\
+      | dftk stat ... -c tip -g bill_bin
 """
 
 

dataframe_toolkit-0.7.0/src/dftk/commands/chi_cmd.py

@@ -0,0 +1,193 @@
+"""
+dftk.commands.chi_cmd — dftk chi subcommand.
+
+Corner-resolved chi-plot: a visual diagnostic for which direction (if any)
+two variables have tail dependence in, complementing the numeric
+lambda_lower/lambda_upper columns from `dftk corr --tail`.
+
+Follows the structure of scat_cmd.py -- same --groupcol / --subgraphcol /
+--size / --fontsize figure conventions as the rest of the Plots group.
+"""
+
+import argparse
+
+import numpy as np
+
+from dftk.commands.base import BaseCommand
+from dftk.common.copula import corner_chi, pseudo_observations
+from dftk.common.io import check_cols, io
+from dftk.common.plot import (
+    add_figure_arguments,
+    add_group_arguments,
+    add_legend_arguments,
+    add_xy_arguments,
+    apply_labels,
+    apply_style,
+    make_figure,
+    make_grouplabel,
+    save_or_show,
+    subgraph_groups,
+    subgraph_layout,
+)
+
+# ---------------------------------------------------------------------------
+# Single-axes chi-plot (reused per subplot)
+# ---------------------------------------------------------------------------
+
+
+def _chi_ax(ax, df, args, title_suffix: str = ""):
+    xcol, ycol = args.xcol, args.ycol
+    legendloc = tuple(args.legendloc) if getattr(args, "legendloc", None) else "best"
+
+    def _draw(sub_df, label=None, **scatter_kw):
+        a, b = sub_df[[xcol, ycol]].dropna().values.T
+        if len(a) < 3:
+            return
+        u, v = pseudo_observations(a, b)
+        x, chi = corner_chi(u, v)
+        mask = np.abs(x) < 0.8  # trim the noisiest extreme points, standard practice
+        ax.scatter(x[mask], chi[mask], s=8, alpha=0.4, label=label, **scatter_kw)
+        ci = 1.78 / np.sqrt(len(u))
+        return ci
+
+    if args.groupcol is not None:
+        # Each group gets its OWN chi statistic computed within its own subset
+        # (mixing groups into one leave-one-out CDF would misrepresent both) --
+        # overlaid here the same way scat/hist overlay per-group series.
+        cis = []
+        for groupname, gdf in subgraph_groups(
+            df, args.groupcol, getattr(args, "groupcolorder", None)
+        ):
+            label = make_grouplabel(
+                groupname, args.groupcol, getattr(args, "groupcolformat", None)
+            )
+            ci = _draw(gdf, label=label)
+            if ci is not None:
+                cis.append(ci)
+        ax.legend(loc=legendloc, title=getattr(args, "legendtitle", None))
+        ci = max(cis) if cis else 0.0
+    else:
+        ci = _draw(df) or 0.0
+
+    ax.axhline(0, color="black", lw=0.8)
+    ax.axvline(0, color="gray", lw=0.6, ls=":")
+    if ci:
+        ax.axhline(ci, color="crimson", lw=1, ls="--", alpha=0.7)
+        ax.axhline(-ci, color="crimson", lw=1, ls="--", alpha=0.7)
+    ax.set_xlim(-1, 1)
+    ax.set_ylim(-0.2, 1)
+
+    apply_labels(
+        ax,
+        args,
+        default_xlabel="<- lower corner    |    upper corner ->",
+        default_ylabel=r"$\chi$ (local dependence)",
+    )
+    if title_suffix:
+        existing = ax.get_title()
+        ax.set_title(f"{existing}  [{title_suffix}]" if existing else title_suffix)
+
+    ax.grid(True, linewidth=0.4, alpha=0.5)
+
+
+# ---------------------------------------------------------------------------
+# Command
+# ---------------------------------------------------------------------------
+
+_EPILOG = """\
+Reads as: chi elevated (and rising) on the LEFT (negative x) only => lower-
+tail dependence -- the two variables' worst values tend to co-occur, but
+their best values don't (e.g. joint crashes but not joint rallies). Elevated
+on the RIGHT only => upper-tail dependence (joint highs, not joint lows).
+Elevated similarly on BOTH sides, decaying back toward 0 well before the
+edges => generic correlation with no real tail dependence. Flat at 0
+everywhere => independence. Dashed red lines mark an approximate 95%
+control limit for "indistinguishable from independence".
+
+This is a visual complement to `dftk corr --tail`'s numeric
+lambda_lower/lambda_upper columns -- same underlying question, this command
+answers it by eye and per-corner in one plot instead of at a chosen set of
+thresholds. Needs a reasonable amount of data to be legible: a few hundred
+points is usually enough to see the shape, but a strong one-sided asymmetry
+is easiest to trust with several thousand.
+
+EXAMPLES
+--------
+Basic chi-plot:
+
+  dftk chi data.tsv -x recomb_rate -y gc_content
+  dftk dataset tips -o | dftk chi ... -x total_bill -y tip
+
+Split into a subplot grid by category, to check whether tail dependence is
+consistent across the whole dataset or driven by a few subgroups (e.g.
+chromosome, sector, region — here, diamond cut quality):
+
+  dftk dataset diamonds -o | dftk chi ... -x carat -y price --subgraphcol cut
+
+Colour-code by group instead of splitting into subplots:
+
+  dftk dataset diamonds -o | dftk chi ... -x carat -y price -g cut
+
+Publication figure (Nature single column, PDF):
+
+  dftk dataset tips -o | dftk chi ... -x total_bill -y tip \\
+      --size single --fontsize publication -f fig.pdf
+"""
+
+
+class ChiCommand(BaseCommand):
+    name = "chi"
+    help = "Corner-resolved chi-plot: visualize lower vs upper tail dependence."
+
+    def add_arguments(self, parser: argparse.ArgumentParser) -> None:
+        parser.formatter_class = argparse.RawDescriptionHelpFormatter
+        parser.epilog = _EPILOG
+
+        self.add_read_arguments(parser)
+        add_xy_arguments(parser)
+        add_figure_arguments(parser)
+        add_group_arguments(parser)
+        add_legend_arguments(parser)
+
+    def execute(self, args: argparse.Namespace) -> None:
+        if not args.xcol or not args.ycol:
+            raise ValueError("--xcol (-x) and --ycol (-y) are required.")
+
+        df = io.read(args)
+        check_cols(df, [args.xcol, args.ycol], "-x/-y")
+        check_cols(df, args.groupcol, "-g/--groupcol")
+        check_cols(df, args.subgraphcol, "--subgraphcol")
+
+        if args.file:
+            import matplotlib
+
+            matplotlib.use("Agg")
+
+        import matplotlib.pyplot as plt
+
+        if getattr(args, "usetex", False):
+            plt.rc("text", usetex=True)
+
+        apply_style(args)
+
+        if args.subgraphcol is not None:
+            groups = subgraph_groups(
+                df, args.subgraphcol, getattr(args, "subgraphorder", None)
+            )
+            nrows, ncols = subgraph_layout(len(groups), args.ncols)
+            fig, axes = make_figure(args, nrows=nrows, ncols=ncols)
+            for idx, (groupname, gdf) in enumerate(groups):
+                ax = axes[idx // ncols][idx % ncols]
+                suffix = make_grouplabel(
+                    groupname, args.subgraphcol, getattr(args, "subgraphformat", None)
+                )
+                _chi_ax(ax, gdf, args, title_suffix=suffix)
+            for idx in range(len(groups), nrows * ncols):
+                axes[idx // ncols][idx % ncols].set_visible(False)
+            fig.tight_layout()
+        else:
+            fig, axes = make_figure(args)
+            _chi_ax(axes[0][0], df, args)
+            fig.tight_layout()
+
+        save_or_show(fig, args)

{dataframe_toolkit-0.5.1 → dataframe_toolkit-0.7.0}/src/dftk/commands/concat_cmd.py

@@ -8,9 +8,11 @@ present in some but not all files are filled with NaN (or --fill value).
 Examples
 --------
     dftk concat a.tsv b.tsv c.tsv
-    dftk concat a.tsv b.tsv --sourcecol source
-    dftk concat a.parquet b.parquet -o
-    dftk eval data.tsv -f "x=1" -o | dftk concat ... extra.tsv
+    dftk dataset penguins -o penguins.tsv
+    dftk concat penguins.tsv penguins.tsv --sourcecol source
+    dftk dataset penguins -o penguins.parquet
+    dftk concat penguins.parquet penguins.parquet -o
+    dftk dataset penguins -o | dftk eval ... -f "x=1" -o | dftk concat ... penguins.tsv
 """
 
 import argparse
@@ -43,9 +45,16 @@ COLUMN ALIGNMENT
 EXAMPLES
 --------
   dftk concat a.tsv b.tsv
-  dftk concat a.tsv b.tsv c.tsv --sourcecol source
-  dftk concat a.parquet b.parquet -o result.parquet
-  dftk eval data.tsv -f "x=1" -o | dftk concat ... extra.tsv
+  dftk dataset penguins -o penguins.tsv
+  dftk concat penguins.tsv penguins.tsv --sourcecol source
+  dftk dataset penguins -o penguins.parquet
+  dftk concat penguins.parquet penguins.parquet -o result.parquet
+  dftk dataset penguins -o | dftk eval ... -f "x=1" -o | dftk concat ... penguins.tsv
+
+Mismatched columns are outer-joined and filled (NaN by default, or --fill):
+
+  dftk dataset iris -o iris.tsv
+  dftk concat iris.tsv penguins.tsv --fill 0
 """
         g = parser.add_argument_group("concat options")
         g.add_argument(
@@ -86,12 +95,6 @@ EXAMPLES
             help="Input files have no header row; columns are named V1, V2, …",
             action="store_true",
         )
-        r.add_argument(
-            "--nrows",
-            help="Maximum number of rows to read per file",
-            type=int,
-            default=None,
-        )
         r.add_argument(
             "--delimiter",
             help="Column delimiter for TSV inputs (default: tab)",

{dataframe_toolkit-0.5.1 → dataframe_toolkit-0.7.0}/src/dftk/commands/corr_cmd.py

@@ -18,6 +18,11 @@ import pandas as pd
 import scipy.stats as ss
 
 from dftk.commands.base import BaseCommand
+from dftk.common.copula import (
+    MIN_TAIL_N,
+    empirical_tail_dependence,
+    pseudo_observations,
+)
 from dftk.common.io import check_cols, io
 from dftk.common.seed import normalize_seed
 
@@ -112,6 +117,29 @@ def _bootstrap_stats(
     return p_perm, ci_lo, ci_hi
 
 
+def _tail_stats(a: np.ndarray, b: np.ndarray, q_grid: np.ndarray) -> tuple:
+    """Return a flat tuple of (lambda_lower_qX, lambda_upper_qX, ...) for each q
+    in q_grid, in the same order, plus a stderr warning if any q is too small
+    for the sample size to resolve.
+    """
+    n = len(a)
+    u, v = pseudo_observations(a, b)
+    lam_L, lam_U = empirical_tail_dependence(u, v, q_grid)
+    thin = [q for q in q_grid if n * q < MIN_TAIL_N]
+    if thin:
+        logger.warning(
+            "tail dependence: n=%d gives fewer than %d points in the tail for "
+            "q=%s -- these estimates will be noisy. See --tail-q / MIN_TAIL_N.",
+            n,
+            MIN_TAIL_N,
+            ", ".join(f"{q:g}" for q in thin),
+        )
+    out = []
+    for lo, hi in zip(lam_L, lam_U, strict=True):
+        out.extend([float(lo), float(hi)])
+    return tuple(out)
+
+
 def _corr(
     df: pd.DataFrame, args: argparse.Namespace, rng: np.random.Generator
 ) -> pd.DataFrame:
@@ -122,12 +150,17 @@ def _corr(
         raise ValueError(f"-c pairs must be col1:col2, got: {bad}")
 
     ci = args.ci and args.method == "pearson"
+    tail = getattr(args, "tail", False)
+    q_grid = np.asarray(getattr(args, "tail_q", None) or [0.05])
 
     header = ["col1", "col2", "nobs", "correlation", "pvalue"]
     if ci:
         header += ["cilo", "cihi"]
     if args.bootstrap is not None:
         header += ["p_perm", "ci_boot_lo", "ci_boot_hi"]
+    if tail:
+        for q in q_grid:
+            header += [f"lambda_lower_q{q:g}", f"lambda_upper_q{q:g}"]
 
     rows = []
 
@@ -136,13 +169,14 @@ def _corr(
             pair_df = sub[[col1, col2]].dropna()
             a, b = pair_df[col1].values, pair_df[col2].values
             cr = _compute_correlation(a, b, method_fn, ci)
-            if args.bootstrap is None:
-                rows.append((*prefix, col1, col2, *cr))
-            else:
-                boot = _bootstrap_stats(
+            extra = ()
+            if args.bootstrap is not None:
+                extra += _bootstrap_stats(
                     a, b, method_fn, args.bootstrap, rng, args.confidence / 100.0
                 )
-                rows.append((*prefix, col1, col2, *cr, *boot))
+            if tail:
+                extra += _tail_stats(a, b, q_grid)
+            rows.append((*prefix, col1, col2, *cr, *extra))
 
     if args.groups:
         header = [*args.groups, *header]
@@ -161,7 +195,7 @@ COLUMN PAIRS (-c)
   Pairs are specified as col1:col2.  Multiple pairs may be given:
 
     dftk corr data.tsv -c x:y
-    dftk corr data.tsv -c x:y a:b
+    dftk dataset tips -o | dftk corr ... -c total_bill:tip tip:size
 
 METHODS (--method)
 ------------------
@@ -195,12 +229,45 @@ BOOTSTRAP (--bootstrap N)
 
   Use --randomseed for reproducibility.
 
+TAIL DEPENDENCE (--tail / --tail-q)
+------------------------------------
+  Adds lambda_lower_qN / lambda_upper_qN columns: nonparametric estimates of
+  how likely one variable is to be extreme given the other is, separately
+  for the lower and upper tail, at each threshold N given to --tail-q.
+
+  A correlation coefficient describes overall association; it says nothing
+  about whether the *extremes* specifically move together more (or less)
+  than the bulk of the data. Two pairs can share the same Spearman rho and
+  have very different tail behavior -- this is what tells them apart. Two
+  markers can look equally correlated genome-wide while only one shares
+  its rare, extreme values; two assets can look equally correlated day to
+  day while only one actually crashes together.
+
+  Reading it: pass 2-3 values to --tail-q spanning an order of magnitude
+  (e.g. --tail-q 0.02 0.05 0.1). If both lambda columns shrink as q shrinks,
+  there is no real tail dependence (whatever correlation exists is spread
+  evenly, not concentrated in the extremes). If a lambda column stays
+  roughly flat across q instead of shrinking, that side has genuine tail
+  dependence. A single q in isolation is easy to misread, since even a pair
+  with NO tail dependence gives an elevated, nonzero reading at any finite
+  q if the pair is strongly correlated overall -- it is the trend across q
+  that carries the signal, not any one number.
+
+  This needs enough data in the tail region to be reliable: a warning is
+  printed to stderr when nobs * q for the smallest requested q falls below
+  20, since anything thinner is dominated by sampling noise. Pair with
+  `dftk chi` for a visual, corner-resolved view of the same question.
+
 EXAMPLES
 --------
   dftk corr data.tsv -c x:y
-  dftk corr data.tsv -c x:y -g group --method spearman
-  dftk corr data.tsv -c x:y --ci
-  dftk corr data.tsv -c x:y --method spearman --bootstrap 2000 --randomseed 42
+  dftk dataset tips -o | dftk corr ... -c total_bill:tip
+  dftk dataset tips -o | dftk corr ... -c total_bill:tip -g day --method spearman
+  dftk dataset tips -o | dftk corr ... -c total_bill:tip --ci
+  dftk dataset tips -o \\
+      | dftk corr ... -c total_bill:tip --method spearman \\
+            --bootstrap 2000 --randomseed 42
+  dftk dataset diamonds -o | dftk corr ... -c carat:price --tail --tail-q 0.02 0.05 0.1
 """
 
 
@@ -267,6 +334,22 @@ class CorrCommand(BaseCommand):
             metavar="SEED",
             help="Random seed for bootstrap/permutation (integer or string).",
         )
+        g.add_argument(
+            "--tail",
+            action="store_true",
+            help="Add nonparametric lower/upper tail-dependence coefficient "
+            "columns (see --tail-q).",
+        )
+        g.add_argument(
+            "--tail-q",
+            type=float,
+            nargs="+",
+            default=None,
+            metavar="Q",
+            help="Tail probability threshold(s) for --tail, e.g. 0.02 0.05 0.1 "
+            "(default: 0.05). Smaller q looks deeper into the tail but needs "
+            "more data to stay reliable -- see epilog.",
+        )
 
     def execute(self, args: argparse.Namespace) -> None:
         seed = normalize_seed(args.randomseed) if args.randomseed is not None else None

{dataframe_toolkit-0.5.1 → dataframe_toolkit-0.7.0}/src/dftk/commands/describe_cmd.py

@@ -37,7 +37,7 @@ Optional flags
 Example
 -------
     dftk describe data.tsv
-    dftk describe data.tsv --summary --correlations
+    dftk dataset iris -o | dftk describe ... --summary --correlations
 """
 
 import argparse
@@ -314,9 +314,9 @@ NOTES FLAGS
 EXAMPLES
 --------
   dftk describe data.tsv
-  dftk describe data.tsv --summary --correlations
-  dftk describe data.tsv --summary --correlations --corr-threshold 0.5
-  dftk dataset iris | dftk describe - --summary --correlations
+  dftk dataset iris -o | dftk describe ...
+  dftk dataset iris -o | dftk describe ... --summary --correlations
+  dftk dataset iris -o | dftk describe ... --summary --correlations --corr-threshold 0.5
 """
         io.parser_read(parser)
 

{dataframe_toolkit-0.5.1 → dataframe_toolkit-0.7.0}/src/dftk/commands/eval_cmd.py

@@ -43,7 +43,8 @@ Each -f argument is first passed to pandas df.eval().  This handles
 arithmetic and boolean expressions directly on column names:
 
   dftk eval data.tsv -f "z = x + y * 2"
-  dftk eval data.tsv -f "flag = (pval < 0.05) & (effect > 0)"
+  dftk dataset tips -o | dftk eval ... -f "bill_and_tip = total_bill + tip * 2"
+  dftk dataset tips -o | dftk eval ... -f "good_tip = (tip > 3) & (total_bill < 30)"
 
 If df.eval() fails (e.g. for function calls), the expression is retried
 as a special function call using the syntax:
@@ -61,8 +62,8 @@ Available special functions:
         true_val and false_val can each be: a column name, a quoted
         string literal ("foo"), or a numeric literal.
       Example:
-        dftk eval data.tsv \
-          -f "big = value > 100" \
+        dftk dataset tips -o | dftk eval ... \
+          -f "big = total_bill > 30" \
           -f "label = where(big, \"large\", \"small\")"
 
   Bitwise:
@@ -119,8 +120,8 @@ CONSTANT COLUMNS (-c)
 ----------------------
 Add a column with a fixed value:
 
-  dftk eval data.tsv -c "batch = 3"
-  dftk eval data.tsv -c "label = control"
+  dftk dataset tips -o | dftk eval ... -c "batch = 3"
+  dftk dataset tips -o | dftk eval ... -c "region = northeast"
 
 The value is coerced to numeric when possible.
 

{dataframe_toolkit-0.5.1 → dataframe_toolkit-0.7.0}/src/dftk/commands/fit_cmd.py

@@ -155,26 +155,27 @@ EXAMPLES
 Simple OLS:
 
   dftk fit data.tsv -f "y ~ x"
+  dftk dataset mpg -o | dftk fit ... -f "mpg ~ horsepower"
 
 Grouped OLS (one regression per group):
 
-  dftk fit data.tsv -f "y ~ x" -g group
+  dftk dataset mpg -o | dftk fit ... -f "mpg ~ horsepower" -g origin
 
 Weighted least squares:
 
-  dftk fit data.tsv -f "y ~ x" -w weight_col
+  dftk dataset mpg -o | dftk fit ... -f "mpg ~ horsepower" -w weight
 
 Robust regression (HuberT):
 
-  dftk fit data.tsv -f "y ~ x" --robust
+  dftk dataset mpg -o | dftk fit ... -f "mpg ~ horsepower" --robust
 
 Print detailed summary (no grouping):
 
-  dftk fit data.tsv -f "y ~ x" --summary
+  dftk dataset mpg -o | dftk fit ... -f "mpg ~ horsepower" --summary
 
 ANOVA table (OLS, no grouping):
 
-  dftk fit data.tsv -f "y ~ x + z" --anova
+  dftk dataset mpg -o | dftk fit ... -f "mpg ~ horsepower + weight" --anova
 """