pd.Series(list(a))
is consistently slower than
pd.Series(a.tolist())
tested 20,000,000 -- 500,000 rows
a = np.ones((500000,2))
showing only 1,000,000 rows:
%timeit pd.Series(list(a))
1 loop, best of 3: 301 ms per loop
%timeit pd.Series(a.tolist())
1 loop, best of 3: 261 ms per loop