Distributions: histplot, kdeplot & ecdfplot

The problem, from first principles

Seaborn is matplotlib with the statistics built in. You hand it a table and the names of columns, and it does the grouping, binning and smoothing before drawing. For one numeric column it offers three views of the same distribution: a histogram (counts in bins), a kernel density estimate (a smooth curve that integrates to 1) and an empirical CDF (the fraction of values at or below each point, with no binning at all).

Each is a function that draws on an Axes and returns it, so everything it drew can be read back: the bar heights, the curve, the labels. The details of what scale each one uses (counts, density or proportion) is exactly the sort of thing that changes a picture's meaning, so this question builds all three, plus a comparison of one distribution per group on a fair, per-group scale.

From theory to code

Implement histogram_axes(df, column, bins, stat), kde_axes(df, column), kde_by_group(df, column, group, order) and ecdf_axes(df, column). The signatures and docstrings are already in the editor.

Constraints

• Draw with seaborn (import seaborn as sns), passing data=df and the column names, and return the Axes the call returns. Do not call plt.show().
• histogram_axes(df, column, bins, stat) draws a histogram of column with bins bins and the statistic stat ("count", "density" or "probability").
• kde_axes(df, column) draws one kernel density curve of column.
• kde_by_group(df, column, group, order) draws one density curve per value of group, in the colour order given by order (a list of the group values), each curve normalised on its own (so every curve integrates to 1 whatever the group size).
• ecdf_axes(df, column) draws the empirical cumulative distribution function of column.

Hints