Summarize Named Evaluation Groups

Task

Write summarize_evaluation_groups(expected, produced, row_groups, group_order).

The three row sequences are aligned: item i in each sequence describes one comparison. expected[i] and produced[i] may be strings or numbers, and they match when Python equality says they match. row_groups[i] names the group for that row. group_order is a non-empty sequence of unique group names and defines the order of the output, including groups with no rows. Every row group must occur in group_order.

Return a dictionary with groups, total, correct, accuracy, and error_rate. groups is a list of records in exactly group_order; each record has:

group, count, correct, accuracy, error_rate

For a group, count is its number of rows and correct is the number for which expected equals produced. For a non-empty group:

accuracy  = correct / count
error_rate = (count - correct) / count

For an empty group, both fractions are exactly 0.0. The top-level values use the same rules over all rows; for an input with no rows, total and correct are 0 and both top-level fractions are 0.0.

If the row sequences have different lengths, group_order is empty or has a duplicate, or a row names an unknown group, raise ValueError with "invalid evaluation groups". A rejected input must not be partly returned.

Example

The records are short: (count=3, correct=2), long: (1, 0), and empty: (0, 0). Their accuracies are 2/3, 0.0, and 0.0; the top-level accuracy is 2/4.

Your implementation

Edit solution.py and keep this function signature:

Return ordinary Python numbers in the records. Preserve every input sequence and its row order. Do not print or ask for input.