def render_correlation_impact(itmdt: Intermediate, plot_width: int,
plot_height: int,
palette: Sequence[str]) -> Dict[str, Any]:
"""
Render correlation heatmaps in to tabs
"""
tabs: List[Panel] = []
tooltips = [("x", "@x"), ("y", "@y"),
("correlation", "@correlation{1.11}")]
axis_range = itmdt["axis_range"]
for method, df in itmdt["data"].items():
# in case of numerical column names
df = df.copy()
df["x"] = df["x"].apply(str)
df["y"] = df["y"].apply(str)
mapper, color_bar = create_color_mapper(palette)
x_range = FactorRange(*axis_range)
y_range = FactorRange(*reversed(axis_range))
fig = Figure(
x_range=x_range,
y_range=y_range,
plot_width=plot_width,
plot_height=plot_height,
x_axis_location="below",
tools="hover",
toolbar_location=None,
tooltips=tooltips,
background_fill_color="#fafafa",
title=" ",
)
tweak_figure(fig)
fig.rect(
x="x",
y="y",
width=1,
height=1,
source=df,
fill_color={
"field": "correlation",
"transform": mapper
},
line_color=None,
)
fig.frame_width = plot_width
fig.add_layout(color_bar, "left")
tab = Panel(child=fig, title=method)
tabs.append(tab)
return {
"insights": itmdt["insights"],
"tabledata": itmdt["tabledata"],
"layout": [panel.child for panel in tabs],
"meta": [panel.title for panel in tabs],
"container_width": plot_width + 150,
}
def render_dendrogram(dend: Dict["str", Any], plot_width: int,
plot_height: int) -> Figure:
"""
Render a missing dendrogram.
"""
# list of lists of dcoords and icoords from scipy.dendrogram
xs, ys, cols = dend["icoord"], dend["dcoord"], dend["ivl"]
# if the number of columns is greater than 20, make the plot wider
if len(cols) > 20:
plot_width = 28 * len(cols)
fig = Figure(
plot_width=plot_width,
plot_height=plot_height,
toolbar_location=None,
tools="",
title=" ",
)
# round the coordinates to integers, and plot the dendrogram
xs = [[round(coord) for coord in coords] for coords in xs]
ys = [[round(coord, 2) for coord in coords] for coords in ys]
fig.multi_line(xs=xs, ys=ys, line_color="#8073ac")
# extract the horizontal lines for the hover tooltip
h_lns_x = [coords[1:3] for coords in xs]
h_lns_y = [coords[1:3] for coords in ys]
null_mismatch_vals = [coord[0] for coord in h_lns_y]
source = ColumnDataSource(dict(x=h_lns_x, y=h_lns_y, n=null_mismatch_vals))
h_lns = fig.multi_line(xs="x", ys="y", source=source, line_color="#8073ac")
hover_pts = HoverTool(
renderers=[h_lns],
tooltips=[("Average distance", "@n{0.1f}")],
line_policy="interp",
)
fig.add_tools(hover_pts)
# shorten column labels if necessary, and override coordinates with column names
cols = [f"{col[:16]}..." if len(col) > 18 else col for col in cols]
axis_coords = list(range(5, 10 * len(cols) + 1, 10))
axis_overrides = dict(zip(axis_coords, cols))
fig.xaxis.ticker = axis_coords
fig.xaxis.major_label_overrides = axis_overrides
fig.xaxis.major_label_orientation = np.pi / 3
fig.yaxis.axis_label = "Average Distance Between Clusters"
fig.grid.visible = False
fig.frame_width = plot_width
return fig
def render_bar_chart(
data: Tuple[np.ndarray, np.ndarray, np.ndarray],
yscale: str,
plot_width: int,
plot_height: int,
) -> Figure:
"""
Render a bar chart for the missing and present values
"""
pres_cnts, null_cnts, cols = data
df = pd.DataFrame({"Present": pres_cnts, "Missing": null_cnts}, index=cols)
if len(df) > 20:
plot_width = 28 * len(df)
fig = Figure(
x_range=list(df.index),
y_range=[0, df["Present"][0] + df["Missing"][0]],
plot_width=plot_width,
plot_height=plot_height,
y_axis_type=yscale,
toolbar_location=None,
tools=[],
title=" ",
)
rend = fig.vbar_stack(
stackers=df.columns,
x="index",
width=0.9,
color=[CATEGORY20[0], CATEGORY20[2]],
source=df,
legend_label=list(df.columns),
)
# hover tool with count and percent
formatter = CustomJSHover(
args=dict(source=ColumnDataSource(df)),
code="""
const columns = Object.keys(source.data)
const cur_bar = special_vars.data_x - 0.5
var ttl_bar = 0
for (let i = 0; i < columns.length; i++) {
if (columns[i] != 'index'){
ttl_bar = ttl_bar + source.data[columns[i]][cur_bar]
}
}
const cur_val = source.data[special_vars.name][cur_bar]
return (cur_val/ttl_bar * 100).toFixed(2)+'%';
""",
)
for i, val in enumerate(df.columns):
hover = HoverTool(
tooltips=[
("Column", "@index"),
(f"{val} count", "@$name"),
(f"{val} percent", "@{%s}{custom}" % rend[i].name),
],
formatters={"@{%s}" % rend[i].name: formatter},
renderers=[rend[i]],
)
fig.add_tools(hover)
fig.yaxis.axis_label = "Row Count"
tweak_figure(fig)
relocate_legend(fig, "left")
fig.frame_width = plot_width
return fig
def render_heatmaps(df: Optional[pd.DataFrame], plot_width: int, plot_height: int) -> Figure:
"""
Render missing heatmaps in to tabs
"""
tooltips = [("x", "@x"), ("y", "@y"), ("correlation", "@correlation{1.11}")]
mapper, color_bar = create_color_mapper_heatmap(RDBU)
def empty_figure() -> Figure:
# If no data to render in the heatmap, i.e. no missing values
# we render a blank heatmap
fig = Figure(
x_range=[],
y_range=[],
plot_width=plot_width,
plot_height=plot_height,
x_axis_location="below",
tools="hover",
toolbar_location=None,
background_fill_color="#fafafa",
)
# Add at least one renderer to fig, otherwise bokeh
# gives us error -1000 (MISSING_RENDERERS): Plot has no renderers
fig.rect(x=0, y=0, width=0, height=0)
return fig
if df is not None:
df = df.where(np.triu(np.ones(df.shape)).astype(np.bool)).T # pylint: disable=no-member
if df.size != 0:
x_range = FactorRange(*df.columns)
y_range = FactorRange(*reversed(df.columns))
df = df.unstack().reset_index(name="correlation")
df = df.rename(columns={"level_0": "x", "level_1": "y"})
df = df[df["x"] != df["y"]]
df = drop_null(df)
# in case of numerical column names
df["x"] = df["x"].apply(str)
df["y"] = df["y"].apply(str)
fig = Figure(
x_range=x_range,
y_range=y_range,
plot_width=plot_width,
plot_height=plot_height,
x_axis_location="below",
tools="hover",
toolbar_location=None,
tooltips=tooltips,
background_fill_color="#fafafa",
title=" ",
)
fig.rect(
x="x",
y="y",
width=1,
height=1,
source=df,
fill_color={"field": "correlation", "transform": mapper},
line_color=None,
)
else:
fig = empty_figure()
else:
fig = empty_figure()
tweak_figure(fig)
fig.grid.grid_line_color = None
fig.axis.axis_line_color = None
fig.add_layout(color_bar, "left")
fig.frame_width = plot_width
return fig
