Perform basic matrix calculations using numpy.
Implement matrix_calculations(matrix_a, matrix_b) for two same-sized square matrices. Return a dictionary containing element-wise sum, matrix product, the transpose of matrix_a, and the determinant of matrix_a. All numeric results must be represented as floats. Implement the required behavior without input/output handling.
matrix_a and matrix_b are square lists of lists. Return {"sum": ..., "product": ..., "transpose": ..., "determinant": ...}.
def matrix_calculations(matrix_a, matrix_b):