yuclid 0.1.0__py3-none-any.whl

yuclid/spread.py

@@ -0,0 +1,152 @@
+from yuclid.log import report, LogLevel
+from scipy.stats import norm
+import seaborn as sns
+import pandas as pd
+import numpy as np
+import re
+
+
+available_parametrized = {"sd", "pi", "rsd"}
+available_non_parametrized = {
+    "mad",
+    "range",
+    "iqr",
+}
+available = [f"{name},N" for name in available_parametrized]
+available += [name for name in available_non_parametrized]
+
+
+def mad(y):
+    if not isinstance(y, np.ndarray):
+        y = np.array(y)
+    return np.median(np.abs(y - np.median(y)))
+
+
+def lower(spread_measure):
+    n = re.search(r"\d+(\.\d+)?", spread_measure)
+    if spread_measure.startswith("sd"):
+        coeff = float(n.group())
+        return lambda y: y.mean() - coeff * y.std()
+    elif spread_measure.startswith("pi"):
+        p = float(n.group()) / 100.0
+        return lambda y: np.quantile(y, (1 - p) / 2)
+    elif spread_measure.startswith("rsd"):
+        coeff = float(n.group())
+        return lambda y: y.mean() - coeff * (1 / norm.ppf(0.75)) * mad(y)
+    elif spread_measure == "mad":
+        return lambda y: np.median(y) - mad(y)
+    elif spread_measure == "range":
+        return lambda y: y.min()
+    elif spread_measure == "iqr":
+        return lambda y: np.quantile(y, 0.25)
+
+
+def upper(spread_measure):
+    n = re.search(r"\d+(\.\d+)?", spread_measure)
+    if spread_measure.startswith("sd"):
+        coeff = float(n.group())
+        return lambda y: y.mean() + coeff * y.std()
+    elif spread_measure.startswith("pi"):
+        p = float(n.group()) / 100.0
+        return lambda y: np.quantile(y, (1 + p) / 2)
+    elif spread_measure.startswith("rsd"):
+        coeff = float(n.group())
+        return lambda y: y.mean() + coeff * (1 / norm.ppf(0.75)) * mad(y)
+    elif spread_measure == "mad":
+        return lambda y: np.median(y) + mad(y)
+    elif spread_measure == "range":
+        return lambda y: y.max()
+    elif spread_measure == "iqr":
+        return lambda y: np.quantile(y, 0.75)
+
+
+def assert_validity(spread_measure):
+    if not isinstance(spread_measure, str):
+        report(LogLevel.ERROR, "spread_measure must be a string")
+        return False
+    parts = spread_measure.split(",")
+    if parts[0] not in available_non_parametrized.union(available_parametrized):
+        report(
+            LogLevel.ERROR,
+            f"spread_measure '{spread_measure}' is not available",
+            hint="use one of: " + " - ".join(available),
+        )
+        return False
+    if parts[0] in available_parametrized:
+        if len(parts) < 2:
+            report(
+                LogLevel.ERROR,
+                f"spread_measure '{spread_measure}' is missing a parameter",
+                hint=f"try '{spread_measure},N' where N is a number",
+            )
+            return False
+        n = float(parts[1])
+        if parts[0] == "pi" and n < 0 or n > 100:
+            report(
+                LogLevel.ERROR,
+                "parameter for spread_measure 'pi' must be in [0, 100]",
+            )
+            return False
+        if parts[0] in ["sd", "rsd"] and n <= 0:
+            report(
+                LogLevel.ERROR,
+                "parameter for spread_measure 'sd' and 'rsd' must be positive",
+            )
+            return False
+    if parts[0] in available_non_parametrized:
+        if len(parts) != 1:
+            report(
+                LogLevel.ERROR,
+                f"spread_measure '{spread_measure}' must not have any parameters",
+            )
+            return False
+    return True
+
+
+def draw(
+    ax, spread_measures, df, x, y, z=None, colors=None, palette=None, style="area"
+):
+    alphas = {
+        "area": np.linspace(0.15, 0.05, len(spread_measures)),
+        "bar": np.linspace(0.30, 0.10, len(spread_measures)),
+    }[style]
+
+    x_dom = df[x].unique()
+    z_dom = df[z].unique()
+    groups = df.groupby([z, x])[y]
+
+    if isinstance(palette, str):
+        colors = sns.color_palette(palette)
+        palette = None
+
+    if colors is None and palette is None:
+        raise Exception("Missing color and palette")
+
+    for i, sm in enumerate(spread_measures):
+        ys_lower = groups.apply(lower(sm))
+        ys_upper = groups.apply(upper(sm))
+        for j, z_val in enumerate(z_dom):
+            if colors is not None:
+                color = colors[j % len(colors)]
+            else:
+                color = palette[z_val]
+            y_lower = ys_lower.xs(z_val)
+            y_upper = ys_upper.xs(z_val)
+            if style == "area":
+                ax.fill_between(
+                    x_dom,
+                    y_lower,
+                    y_upper,
+                    interpolate=True,
+                    color=color,
+                    alpha=alphas[i],
+                )
+            elif style == "bar":
+                ax.vlines(
+                    x=x,
+                    ymin=y_lower,
+                    ymax=y_upper,
+                    color=color,
+                    alpha=alphas[i],
+                    linewidth=4,
+                )

yuclid-0.1.0.dist-info/METADATA

@@ -0,0 +1,15 @@
+Metadata-Version: 2.4
+Name: yuclid
+Version: 0.1.0
+Summary: A Python project.
+Author-email: Your Name <your.email@example.com>
+Project-URL: Homepage, https://github.com/fsossai/yuclid
+Requires-Python: >=3.8
+Description-Content-Type: text/markdown
+
+# Yuclid
+
+*Combinatorially explode your experiments*
+
+- **`yuclid run`**: Run experiments with with all combination of registered parameters.
+- **`yuclid plot`**: Interactively visualizes the results produced by `yuclid run`.

yuclid-0.1.0.dist-info/RECORD

@@ -0,0 +1,11 @@
+yuclid/__init__.py,sha256=Pru0BlFBASFCFo7McHdohtKkUtgMPDwbGfyUZlE2_Vw,21
+yuclid/cli.py,sha256=l5WUY6Q6nwg7WRrAAPf5uaspG9zrEPE9BA9v3eYI_vE,6410
+yuclid/log.py,sha256=GR_FVfNroumuonKguAPd6H1rKjxJKRc8tAS2sVNTbzE,1655
+yuclid/plot.py,sha256=RV_bgkFDpOGxw7ankW7QsnBsyrholBtYKKj9jUtBAyM,30836
+yuclid/run.py,sha256=V3tcFHWyXH2fs6aFR6vJi2c5DhDb-zDXJMWp-xUi8DM,44048
+yuclid/spread.py,sha256=4Ci3nsu8n_dhG-AK2IWHKRElQ8oaGdw14LrgNu79biM,4938
+yuclid-0.1.0.dist-info/METADATA,sha256=umoUK82Bh__fTcoLiFS64thJ6d5brdDWHWmkozzA9p8,473
+yuclid-0.1.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+yuclid-0.1.0.dist-info/entry_points.txt,sha256=2AvTtyt5iBnjr6HnjqH_3PeSoq9UzIbT92qivmEbOYA,43
+yuclid-0.1.0.dist-info/top_level.txt,sha256=cL5mb4h_4etwTsqhPvSnoVBXImIzPFGd3rINV1nEjPo,7
+yuclid-0.1.0.dist-info/RECORD,,

yuclid-0.1.0.dist-info/WHEEL

@@ -0,0 +1,5 @@
+Wheel-Version: 1.0
+Generator: setuptools (80.9.0)
+Root-Is-Purelib: true
+Tag: py3-none-any
+

yuclid-0.1.0.dist-info/entry_points.txt

@@ -0,0 +1,2 @@
+[console_scripts]
+yuclid = yuclid.cli:main

yuclid-0.1.0.dist-info/top_level.txt

@@ -0,0 +1 @@
+yuclid