lessPython 0.2.0__py3-none-any.whl → 0.3.1__py3-none-any.whl

lessPy/Correlation.py

@@ -12,7 +12,7 @@
 import numpy as np
 import pandas as pd
 
-from .utils import fmt, get_column, get_option
+from .utils import fmt, get_column, get_option, file_output, show_plots
 
 
 class CorrelationResults:
@@ -110,6 +110,7 @@ def _two_var(x, y, x_name, y_name, method, brief):
         r, p = sps.pearsonr(xk, yk)
         df = n - 2
         t = r * np.sqrt(df / (1 - r ** 2))
+        stat_nm = "t"
         p = 2 * sps.t.sf(abs(t), df)
         z = np.arctanh(r)
         sig = 1 / np.sqrt(n - 3)
@@ -119,17 +120,41 @@ def _two_var(x, y, x_name, y_name, method, brief):
         sym, sym_pop = "r", "Correlation"
         cov = float(np.cov(xk, yk)[0, 1])
     elif method == "spearman":
+        # the statistic cor.test() reports: S, the sum of the squared
+        #   rank differences, with no df. R analog: cr.main.R
+        #   names(ct$statistic)
         res = sps.spearmanr(xk, yk)
         r, p = float(res.statistic), float(res.pvalue)
-        t, df = res.statistic * np.sqrt(
-            (n - 2) / (1 - r ** 2)), n - 2
+        t, df, stat_nm = (n ** 3 - n) * (1 - r) / 6, None, "S"
         method_txt = "Spearman's rank correlation rho"
         sym = sym_pop = "rho"
     else:
+        # cor.test(): T, the count of concordant pairs, with its exact
+        #   p-value for n < 50 and no ties, else z with the variance
+        #   corrected for ties
         res = sps.kendalltau(xk, yk)
-        r, p = float(res.statistic), float(res.pvalue)
-        t = res.statistic
+        r = float(res.statistic)
+        _, xt = np.unique(xk, return_counts=True)
+        _, yt = np.unique(yk, return_counts=True)
+        ties = (xt > 1).any() or (yt > 1).any()
         df = None
+        if n < 50 and not ties:
+            t, stat_nm = (r + 1) * n * (n - 1) / 4, "T"
+            p = float(sps.kendalltau(xk, yk, method="exact").pvalue)
+        else:
+            T0 = n * (n - 1) / 2
+            S = r * np.sqrt((T0 - (xt * (xt - 1)).sum() / 2)
+                            * (T0 - (yt * (yt - 1)).sum() / 2))
+            v0 = n * (n - 1) * (2 * n + 5)
+            vt = (xt * (xt - 1) * (2 * xt + 5)).sum()
+            vu = (yt * (yt - 1) * (2 * yt + 5)).sum()
+            v1 = (xt * (xt - 1)).sum() * (yt * (yt - 1)).sum()
+            v2 = ((xt * (xt - 1) * (xt - 2)).sum()
+                  * (yt * (yt - 1) * (yt - 2)).sum())
+            var_S = ((v0 - vt - vu) / 18 + v1 / (2 * n * (n - 1))
+                     + v2 / (9 * n * (n - 1) * (n - 2)))
+            t, stat_nm = S / np.sqrt(var_S), "z"
+            p = float(2 * sps.norm.sf(abs(t)))
         method_txt = "Kendall's rank correlation tau"
         sym = sym_pop = "tau"
 
@@ -152,9 +177,9 @@ def _two_var(x, y, x_name, y_name, method, brief):
     else:
         L.append(f"Sample Correlation: {sym} = {fmt(r, 3)}")
     L.append("")
-    dfs = "NA" if df is None else str(df)
-    L.append(f"Hypothesis Test of 0 {sym_pop}:  t = {fmt(t, 3)}, "
-             f" df = {dfs},  p-value = {fmt(p, 3)}")
+    dfs = "" if df is None else f",  df = {df}"
+    L.append(f"Hypothesis Test of 0 {sym_pop}:  {stat_nm} = "
+             f"{fmt(t, 3)}{dfs},  p-value = {fmt(p, 3)}")
     if method == "pearson":
         L.append("95% Confidence Interval for Correlation:  "
                  f"{fmt(lb, 3)} to {fmt(ub, 3)}")
@@ -222,13 +247,13 @@ def _matrix(df, miss, show, method, digits_d, heat_map, main):
     if heat_map:
         plots["heatmap"] = _heatmap(crs, main)
     crs.attrs["plots"] = plots
-    return crs
+    return show_plots(crs)
 
 
 def _heatmap(df, main):
     import plotly.graph_objects as go
 
-    from .plotly_utils import plotly_style, to_hex
+    from .plotly_utils import as_plotly_color, plotly_style, to_hex
     labels = list(df.columns)
     Z = df.to_numpy(dtype=float).copy()
     np.fill_diagonal(Z, np.nan)
@@ -239,9 +264,13 @@ def _heatmap(df, main):
         colorbar=dict(title="r")))
     fig.update_yaxes(autorange="reversed")
     fig.update_layout(
-        template=None, paper_bgcolor=to_hex(style["window_fill"]),
+        template=None, paper_bgcolor=as_plotly_color(style["window_fill"]),
         title=dict(text=main or "Correlations", x=0.5,
                    xanchor="center",
                    font=dict(size=round(
                        16 * get_option("main_size", 1)))))
     return fig
+
+
+# pdf_file= and png_file= write the figures, as lessR's pdf_file=
+Correlation = file_output(Correlation)

lessPy/Flows.py

@@ -9,7 +9,7 @@
 import numpy as np
 import pandas as pd
 
-from .plotly_utils import BASE_COLORS, make_trans, to_hex
+from .plotly_utils import as_plotly_color, BASE_COLORS, make_trans, to_hex
 from .utils import get_column, get_option
 
 
@@ -96,7 +96,7 @@ def Flows(value, stage1, stage2, stage3=None, data=None,
         node=dict(
             label=nodes, color=node_color, x=node_x, y=node_y,
             pad=12, thickness=16,
-            line=dict(color=to_hex(border_gray), width=1),
+            line=dict(color=as_plotly_color(border_gray), width=1),
             hovertemplate="%{label}<extra></extra>"),
         link=dict(
             source=src, target=tgt, value=val, color=lcol,
@@ -116,5 +116,5 @@ def Flows(value, stage1, stage2, stage3=None, data=None,
                     r=30, b=30, l=30),
         font=dict(size=round(15 * labels_size)),
         template=None,
-        paper_bgcolor=to_hex(get_option("window_fill", "white")))
+        paper_bgcolor=as_plotly_color(get_option("window_fill", "white")))
     return fig

lessPy/Logit.py

@@ -36,13 +36,13 @@ import pandas as pd
 import plotly.graph_objects as go
 
 from .plotly_utils import (
-    axis_format, axis_num, make_trans, plot_border,
+    as_plotly_color, axis_format, axis_num, make_trans, plot_border,
     plotly_style, to_hex, x_grid)
 from .plt_mat_plotly import scatter_matrix
 from .logit_rmd import logit_rmd
 from .Regression import (
     _expand_indicators, _parse_formula, _prntbl)
-from .utils import fmt, get_column, get_option, pretty
+from .utils import fmt, get_column, get_option, pretty, file_output, show_plots
 
 
 class LogitResults:
@@ -484,7 +484,7 @@ def Logit(my_formula, data=None, filter=None, ref_group=None,
                   list(data.columns), Rmd, Rmd_data, Rmd_format,
                   Rmd_browser, results, explain, interpret, code)
 
-    return out
+    return show_plots(out)
 
 
 def _logit_confuse(lines, y01, fitv, pc, y_name, lv2_txt,
@@ -564,13 +564,13 @@ def _logit_plot(xv, y01, mu, levels, y_name, x_name, pc,
                     color=make_trans(pt_fill,
                                      1 - transparency),
                     opacity=1,
-                    line=dict(color=to_hex(pt_fill),
+                    line=dict(color=as_plotly_color(pt_fill),
                               width=1)),
         hoverinfo="x+y", showlegend=False))
     od = np.argsort(xv, kind="stable")
     fig.add_trace(go.Scatter(
         x=xv[od], y=mu[od], mode="lines",
-        line=dict(color=to_hex(pt_fill), width=2),
+        line=dict(color=as_plotly_color(pt_fill), width=2),
         hoverinfo="skip", showlegend=False))
     axT1 = pretty(float(xv.min()), float(xv.max()))
     axT2 = [0, 0.2, 0.4, 0.6, 0.8, 1]
@@ -579,7 +579,7 @@ def _logit_plot(xv, y01, mu, levels, y_name, x_name, pc,
     ax_y = axis_num(y_lab, axT2,
                     [f"{v:g}" for v in axT2])
     ax_y.update(range=[-0.10, 1.10], showgrid=True,
-                gridcolor=to_hex(style_opts["grid_col"]),
+                gridcolor=as_plotly_color(style_opts["grid_col"]),
                 gridwidth=1, griddash="dot")
     shapes = x_grid(axT1) + plot_border()
     if pc is not None:                 # threshold crosshairs
@@ -588,13 +588,13 @@ def _logit_plot(xv, y01, mu, levels, y_name, x_name, pc,
         shapes.append(dict(
             type="line", xref="paper", yref="y",
             x0=0, x1=1, y0=pc, y1=pc,
-            line=dict(color=to_hex("gray35"), width=0.75,
+            line=dict(color=as_plotly_color("gray35"), width=0.75,
                       dash="dash")))
         if float(xv.min()) <= x_cut <= float(xv.max()):
             shapes.append(dict(
                 type="line", xref="x", yref="paper",
                 x0=x_cut, x1=x_cut, y0=0, y1=1,
-                line=dict(color=to_hex("gray35"), width=0.75,
+                line=dict(color=as_plotly_color("gray35"), width=0.75,
                           dash="dash")))
     anns = []
     if levels is not None:             # right-axis level labels
@@ -605,11 +605,15 @@ def _logit_plot(xv, y01, mu, levels, y_name, x_name, pc,
                 xanchor="left",
                 font=dict(size=round(
                     14 * get_option("axis_size", 0.9)),
-                    color=to_hex(get_option(
+                    color=as_plotly_color(get_option(
                         "axis_color", "black")))))
     fig.update_layout(
         xaxis=ax_x, yaxis=ax_y, shapes=shapes,
         annotations=anns, template=None,
-        plot_bgcolor=to_hex(style_opts["panel_fill"]),
-        paper_bgcolor=to_hex(style_opts["window_fill"]))
+        plot_bgcolor=as_plotly_color(style_opts["panel_fill"]),
+        paper_bgcolor=as_plotly_color(style_opts["window_fill"]))
     return fig
+
+
+# pdf_file= and png_file= write the figures, as lessR's pdf_file=
+Logit = file_output(Logit)

lessPy/Regression.py

@@ -31,8 +31,9 @@
 # Numerics through statsmodels OLS (imported lazily, as
 # plt_forecast does); influence measures from OLSInfluence.
 # Returns a RegressionResults object holding the tables as
-# DataFrames and the plotly figures in .plots — figures are
-# not auto-shown (no R graphics device to open).
+# DataFrames and the plotly figures in .plots. In a notebook the
+# figures also display as the analysis runs, as lessR draws them;
+# in a script they are not shown (utils.show_plots).
 
 import math
 import re
@@ -49,7 +50,8 @@ from .plotly_utils import (
     BASE_COLORS, as_plotly_color, axis_format, axis_num,
     make_trans, plot_border, plotly_style, to_hex, x_grid)
 from .utils import (
-    category_order, fmt, get_column, get_option, pretty)
+    category_order, fmt, get_column, get_option, pretty,
+    file_output, show_plots)
 from .X import _breaks_from_args
 
 
@@ -1001,7 +1003,7 @@ def Regression(my_formula, data=None, filter=None, digits_d=None,
                 Rmd_browser, results, explain, interpret, code,
                 n_res_rows, n_pred_rows, res_sort, d)
 
-    return out
+    return show_plots(out)
 
 
 def _reg_kfold(yv, Xd, y_name, pred_names, n_keep, formula,
@@ -1226,13 +1228,13 @@ def _moderation_plot(fit, Xd, y_name, mod_info, d):
     ax_x = axis_num(x_name, axT1, axis_format(axT1, d))
     ax_y = axis_num(y_name, axT2, axis_format(axT2, d))
     ax_y.update(showgrid=True,
-                gridcolor=to_hex(style_opts["grid_col"]),
+                gridcolor=as_plotly_color(style_opts["grid_col"]),
                 gridwidth=1, griddash="dot")
     fig.update_layout(
         xaxis=ax_x, yaxis=ax_y,
         shapes=x_grid(axT1) + plot_border(), template=None,
-        plot_bgcolor=to_hex(style_opts["panel_fill"]),
-        paper_bgcolor=to_hex(style_opts["window_fill"]),
+        plot_bgcolor=as_plotly_color(style_opts["panel_fill"]),
+        paper_bgcolor=as_plotly_color(style_opts["window_fill"]),
         legend=dict(title=dict(text=w_name)),
         title=dict(text="Moderator Variable Interaction Plot",
                    x=0.5, xanchor="center",
@@ -1329,13 +1331,13 @@ def _reg_ancova_plot(fit, Xd, yv, y_name, info, d):
     ax_x = axis_num(cov, axT1, axis_format(axT1, d))
     ax_y = axis_num(y_name, axT2, axis_format(axT2, d))
     ax_y.update(showgrid=True,
-                gridcolor=to_hex(style_opts["grid_col"]),
+                gridcolor=as_plotly_color(style_opts["grid_col"]),
                 gridwidth=1, griddash="dot")
     fig.update_layout(
         xaxis=ax_x, yaxis=ax_y,
         shapes=x_grid(axT1) + plot_border(), template=None,
-        plot_bgcolor=to_hex(style_opts["panel_fill"]),
-        paper_bgcolor=to_hex(style_opts["window_fill"]),
+        plot_bgcolor=as_plotly_color(style_opts["panel_fill"]),
+        paper_bgcolor=as_plotly_color(style_opts["window_fill"]),
         legend=dict(title=dict(text=fac)),
         title=dict(text="Scatterplot and Least-Squares Lines",
                    x=0.5, xanchor="center",
@@ -1382,11 +1384,11 @@ def _reg_scatter(xv, yv, x_name, y_name, fit, smapi, digits_d,
                     sizemode="diameter",
                     color=make_trans(pt_fill, 0.9),
                     opacity=1,
-                    line=dict(color=to_hex(pt_fill), width=1)),
+                    line=dict(color=as_plotly_color(pt_fill), width=1)),
         hoverinfo="x+y", showlegend=False))
     fig.add_trace(go.Scatter(
         x=xs, y=sf["mean"].to_numpy(), mode="lines",
-        line=dict(color=to_hex(get_option("fit_color",
+        line=dict(color=as_plotly_color(get_option("fit_color",
                                           "#5C4032")),
                   width=get_option("fit_lwd", 2)),
         hoverinfo="skip", showlegend=False))
@@ -1403,7 +1405,7 @@ def _reg_scatter(xv, yv, x_name, y_name, fit, smapi, digits_d,
     ax_y = axis_num(y_name, axT2,
                     axis_format(axT2, digits_d))
     ax_y.update(showgrid=True,
-                gridcolor=to_hex(style_opts["grid_col"]),
+                gridcolor=as_plotly_color(style_opts["grid_col"]),
                 gridwidth=1, griddash="dot")
     title = ("Reg Line, Confidence & Prediction Intervals"
              if with_bands
@@ -1411,8 +1413,8 @@ def _reg_scatter(xv, yv, x_name, y_name, fit, smapi, digits_d,
     fig.update_layout(
         xaxis=ax_x, yaxis=ax_y,
         shapes=x_grid(axT1) + plot_border(), template=None,
-        plot_bgcolor=to_hex(style_opts["panel_fill"]),
-        paper_bgcolor=to_hex(style_opts["window_fill"]),
+        plot_bgcolor=as_plotly_color(style_opts["panel_fill"]),
+        paper_bgcolor=as_plotly_color(style_opts["window_fill"]),
         title=dict(text=title, x=0.5, xanchor="center",
                    font=dict(size=round(
                        16 * get_option("main_size", 1)))))
@@ -1442,14 +1444,14 @@ def _reg_resfit(fitted, resid, cooks, cooks_cut, labels,
                     sizemode="diameter",
                     color=make_trans(pt_fill, 0.9),
                     opacity=1,
-                    line=dict(color=to_hex(pt_fill), width=0.5)),
+                    line=dict(color=as_plotly_color(pt_fill), width=0.5)),
         hoverinfo="x+y", showlegend=False))
     flag = np.where(cooks >= cut)[0]
     for i in flag:
         fig.add_annotation(
             x=float(fitted[i]), y=float(resid[i]),
             text=str(labels[i]), showarrow=False, yshift=12,
-            font=dict(size=11, color=to_hex("gray30")))
+            font=dict(size=11, color=as_plotly_color("gray30")))
     axT1 = pretty(float(fitted.min()), float(fitted.max()))
     axT2 = pretty(float(resid.min()), float(resid.max()))
     ax_x = axis_num("Fitted Values", axT1,
@@ -1457,18 +1459,18 @@ def _reg_resfit(fitted, resid, cooks, cooks_cut, labels,
     ax_y = axis_num("Residuals", axT2,
                     axis_format(axT2, digits_d))
     ax_y.update(showgrid=True,
-                gridcolor=to_hex(style_opts["grid_col"]),
+                gridcolor=as_plotly_color(style_opts["grid_col"]),
                 gridwidth=1, griddash="dot")
     shapes = x_grid(axT1) + plot_border()
     shapes.append(dict(                # zero residual line
         type="line", xref="paper", yref="y",
         x0=0, x1=1, y0=0, y1=0,
-        line=dict(color=to_hex("gray50"), width=1,
+        line=dict(color=as_plotly_color("gray50"), width=1,
                   dash="dash")))
     fig.update_layout(
         xaxis=ax_x, yaxis=ax_y, shapes=shapes, template=None,
-        plot_bgcolor=to_hex(style_opts["panel_fill"]),
-        paper_bgcolor=to_hex(style_opts["window_fill"]),
+        plot_bgcolor=as_plotly_color(style_opts["panel_fill"]),
+        paper_bgcolor=as_plotly_color(style_opts["window_fill"]),
         title=dict(
             text=("Residuals vs Fitted Values<br>"
                   f"<sup>{sub}</sup>"),
@@ -1489,3 +1491,7 @@ def _reg_dn_residual(resid):
                     show_histogram=True, hist_edges=edges,
                     main="Distribution of Residuals")
     return fig
+
+
+# pdf_file= and png_file= write the figures, as lessR's pdf_file=
+Regression = file_output(Regression)