Problem

In a BCG client analytics workflow, raw records often contain missing values. Implement a function that cleans a 2D dataset by either dropping rows with too many missing entries or filling missing entries using a provided default value.

A missing entry is represented by None. Given a list of rows, an integer max_missing, and a fill_value, return a new list of rows after applying these rules:

1. If a row contains more than max_missing missing entries, discard the entire row.
2. Otherwise, replace every None in that row with fill_value.
3. Preserve the original order of the remaining rows.

Input / Output

• Input: records as a list of lists, max_missing as an integer, and fill_value as an integer
• Output: A new list of lists representing the cleaned dataset

Examples

Example 1

• Input: records = [[1, None, 3], [None, None, 5], [7, 8, 9]], max_missing = 1, fill_value = 0
• Output: [[1, 0, 3], [7, 8, 9]]
• Explanation: The second row has 2 missing values, so it is removed. The first row is kept and its missing value is filled with 0.

Example 2

• Input: records = [[None], [4], [None, None]], max_missing = 1, fill_value = -1
• Output: [[-1], [4]]
• Explanation: The last row is dropped because it has 2 missing values. The first row is kept and filled.

Constraints

• 1 <= len(records) <= 10^4
• 1 <= len(records[i]) <= 10^3
• All non-missing values and fill_value are integers
• Missing values are represented only by None