Series, Index Alignment & Arithmetic

The problem, from first principles

A NumPy array knows its values and nothing else. A pandas Series carries a second thing with it, an index of labels, and that changes what arithmetic means. Add two arrays and the element at position 0 meets the element at position 0. Add two Series and the element labelled "mon" meets the element labelled "mon", wherever each one happens to sit. This is called index alignment, and it is the reason a pandas calculation can silently give a different answer than the same numbers in an array: if a label appears on only one side, there is nothing to pair it with.

This question builds the four small operations that make alignment concrete: creating a labelled Series, adding two Series by label (with and without a fill value), turning values into shares of a total, and looking labels up without crashing on one that is not there.

From theory to code

Implement make_series(values, labels, name), aligned_sum(a, b, fill_value=None), share_of_total(s) and lookup(s, labels). The signatures and docstrings are already in the editor.

Constraints

• labels are unique. values and labels have the same length.
• make_series returns a pd.Series with those values, that index (in the given order) and that name.
• aligned_sum(a, b) adds a and b by label. The result's index holds every label that appears in a or in b, sorted ascending. A label missing from one side gives NaN.
• aligned_sum(a, b, fill_value=v) treats a missing label or a NaN value on either side as v before adding. If both sides are missing or NaN for a label, the result for that label stays NaN.
• aligned_sum never changes a or b.
• share_of_total(s) returns a float Series with the same index and name as s, where each value is divided by the sum of the non-missing values. Missing values stay missing. The non-missing values do not sum to zero.
• lookup(s, labels) returns a Series indexed by labels, in that order, holding s's value for each label and NaN for a label s does not have.

Hints