Create mock data and plot

dp_creator_ii/app/analysis_panel.py

@@ -1,16 +1,34 @@
-from shiny import ui, reactive
+from shiny import ui, reactive, render
+
+from dp_creator_ii.mock_data import mock_data, ColumnDef
+from dp_creator_ii.app.plots import plot_error_bars_with_cutoff
 
 
 def analysis_ui():
     return ui.nav_panel(
         "Define Analysis",
         "TODO: Define analysis",
+        ui.output_plot("plot_preview"),
+        "(This plot is only to demonstrate that plotting works.)",
         ui.input_action_button("go_to_results", "Download results"),
         value="analysis_panel",
     )
 
 
 def analysis_server(input, output, session):
+    @render.plot()
+    def plot_preview():
+        min_x = 0
+        max_x = 100
+        df = mock_data({"col_0_100": ColumnDef(min_x, max_x)}, row_count=20)
+        return plot_error_bars_with_cutoff(
+            df["col_0_100"].to_list(),
+            x_min_label=min_x,
+            x_max_label=max_x,
+            y_cutoff=30,
+            y_error=5,
+        )
+
     @reactive.effect
     @reactive.event(input.go_to_results)
     def go_to_results():

dp_creator_ii/app/plots.py

@@ -0,0 +1,40 @@
+import matplotlib.pyplot as plt
+import numpy as np
+
+
+def plot_error_bars_with_cutoff(
+    y_values, x_min_label="min", x_max_label="max", y_cutoff=0, y_error=0
+):
+    x_values = 0.5 + np.arange(len(y_values))
+    x_values_above = []
+    x_values_below = []
+    y_values_above = []
+    y_values_below = []
+    for x, y in zip(x_values, y_values):
+        if y < y_cutoff:
+            x_values_below.append(x)
+            y_values_below.append(y)
+        else:
+            x_values_above.append(x)
+            y_values_above.append(y)
+
+    figure, axes = plt.subplots()
+    color = "skyblue"
+    shared = {
+        "width": 0.8,
+        "edgecolor": color,
+        "linewidth": 1,
+        "yerr": y_error,
+    }
+    axes.bar(x_values_above, y_values_above, color=color, **shared)
+    axes.bar(x_values_below, y_values_below, color="white", **shared)
+    axes.hlines([y_cutoff], 0, len(y_values), colors=["black"], linestyles=["dotted"])
+
+    axes.set(xlim=(0, len(y_values)), ylim=(0, max(y_values)))
+    axes.get_xaxis().set_ticks(
+        ticks=[x_values[0], x_values[-1]],
+        labels=[x_min_label, x_max_label],
+    )
+    axes.get_yaxis().set_ticks([])
+
+    return figure

dp_creator_ii/mock_data.py

@@ -0,0 +1,28 @@
+from collections import namedtuple
+import polars as pl
+from scipy.stats import norm  # type: ignore
+
+ColumnDef = namedtuple("ColumnDef", ["min", "max"])
+
+
+def mock_data(column_defs, row_count=1000):
+    schema = {column_name: float for column_name in column_defs.keys()}
+    data = {column_name: [] for column_name in column_defs.keys()}
+
+    # The details here don't really matter: Any method that
+    # deterministically gave us more values in the middle of the range
+    # and fewer at the extremes would do.
+    quantile_width = 95 / 100
+    for column_name, column_def in column_defs.items():
+        min_ppf = norm.ppf((1 - quantile_width) / 2)
+        max_ppf = norm.ppf(1 - (1 - quantile_width) / 2)
+        min_value = column_def.min
+        max_value = column_def.max
+        slope = (max_value - min_value) / (max_ppf - min_ppf)
+        intercept = min_value - slope * min_ppf
+        for i in range(row_count):
+            quantile = (quantile_width * i / (row_count - 1)) + (1 - quantile_width) / 2
+            ppf = norm.ppf(quantile)
+            value = slope * ppf + intercept
+            data[column_name].append(value)
+    return pl.DataFrame(data=data, schema=schema)

dp_creator_ii/tests/test_mock_data.py

@@ -0,0 +1,15 @@
+import polars as pl
+
+from dp_creator_ii.mock_data import mock_data, ColumnDef
+
+
+def test_mock_data():
+    col_0_100 = ColumnDef(0, 100)
+    col_neg_pos = ColumnDef(-10, 10)
+    df = mock_data({"col_0_100": col_0_100, "col_neg_pos": col_neg_pos})
+
+    assert df.select(pl.len()).item() == 1000
+    assert df.get_column("col_0_100")[0] == 0
+    assert df.get_column("col_0_100")[999] == 100
+    assert df.get_column("col_neg_pos")[0] == -10
+    assert df.get_column("col_neg_pos")[999] == 10

pyproject.toml

@@ -12,6 +12,7 @@ dynamic = ["version", "description"]
 dependencies = [
     "shiny",
     "shinywidgets",
+    "matplotlib",
     "opendp[polars]",
     "jupytext",
     "jupyter-client",

requirements-dev.in

@@ -35,3 +35,6 @@ ipykernel
 # Shiny:
 shiny
 shinywidgets
+
+# Visualization:
+matplotlib

requirements-dev.txt

@@ -45,8 +45,12 @@ comm==0.2.2
     # via
     #   ipykernel
     #   ipywidgets
+contourpy==1.3.0
+    # via matplotlib
 coverage==7.6.1
     # via -r requirements-dev.in
+cycler==0.12.1
+    # via matplotlib
 debugpy==1.8.6
     # via ipykernel
 decorator==5.1.1
@@ -73,6 +77,8 @@ flit==3.9.0
     # via -r requirements-dev.in
 flit-core==3.9.0
     # via flit
+fonttools==4.54.1
+    # via matplotlib
 greenlet==3.0.3
     # via playwright
 h11==0.14.0
@@ -124,6 +130,8 @@ jupyterlab-widgets==3.0.13
     # via ipywidgets
 jupytext==1.16.4
     # via -r requirements-dev.in
+kiwisolver==1.4.7
+    # via matplotlib
 linkify-it-py==2.0.3
     # via shiny
 markdown-it-py==3.0.0
@@ -135,6 +143,8 @@ markupsafe==2.1.5
     # via
     #   jinja2
     #   nbconvert
+matplotlib==3.9.2
+    # via -r requirements-dev.in
 matplotlib-inline==0.1.7
     # via
     #   ipykernel
@@ -170,6 +180,8 @@ nodeenv==1.9.1
     # via pre-commit
 numpy==1.26.4
     # via
+    #   contourpy
+    #   matplotlib
     #   opendp
     #   pyarrow
     #   randomgen
@@ -184,6 +196,7 @@ packaging==24.1
     #   htmltools
     #   ipykernel
     #   jupytext
+    #   matplotlib
     #   nbconvert
     #   pytest
     #   shiny
@@ -195,6 +208,8 @@ pathspec==0.12.1
     # via black
 pexpect==4.9.0
     # via ipython
+pillow==10.4.0
+    # via matplotlib
 pip-tools==7.4.1
     # via -r requirements-dev.in
 platformdirs==4.3.6
@@ -233,6 +248,8 @@ pygments==2.18.0
     # via
     #   ipython
     #   nbconvert
+pyparsing==3.1.4
+    # via matplotlib
 pyproject-hooks==1.1.0
     # via
     #   build
@@ -249,6 +266,7 @@ pytest-playwright==0.5.2
 python-dateutil==2.9.0.post0
     # via
     #   jupyter-client
+    #   matplotlib
     #   shinywidgets
 python-multipart==0.0.9
     # via shiny