Missing Data and Feature Prep | Dataford Interview Questions

Write a script to handle missing data and prepare features for a model.

Implement preprocess_rows(rows, numeric_columns, categorical_columns). Each row is a dictionary; missing values are represented by None. Return a dictionary containing features, a numeric matrix, and feature_names, whose order must be consistent across all rows. Impute numeric values with the column median and categorical values with the most frequent non-missing value, then one-hot encode categorical columns using sorted category names. If a column has no observed value, use 0 for numeric imputation and "missing" for categorical imputation.

Constraints

1 <= len(rows) <= 10^4
1 <= len(numeric_columns) + len(categorical_columns) <= 20
Numeric values are integers or floating-point numbers, or None
Categorical values are strings or None
Column names listed in the input arrays exist in every row
Numeric and categorical column lists do not overlap

Function Signature

def preprocess_rows(rows, numeric_columns, categorical_columns):

Your question is Missing Data and Feature Prep. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.

Problem

Write a script to handle missing data and prepare features for a model.

Constraints

1 <= len(rows) <= 10^4
1 <= len(numeric_columns) + len(categorical_columns) <= 20
Numeric values are integers or floating-point numbers, or None
Categorical values are strings or None
Column names listed in the input arrays exist in every row
Numeric and categorical column lists do not overlap

Function Signature

def preprocess_rows(rows, numeric_columns, categorical_columns):

Your question is Missing Data and Feature Prep. Start with the requirements in the Question tab.

Run and submit as often as you like. When you're ready, talk me through your approach or go straight to the code.