DataFrame to list of list without change in data type of values

Question

df.values.to_list() or list(df.values) converts dataframe to list of lists, but the integer values are converted to float values

DataFrame is,

   HSCode  value  year
0       2   0.18  2018
1       3   0.00  2018
2       4  12.48  2018
3       6   0.00  2018
4       7   1.89  2018

output required is

[[2,0.18,2018],[3,0.00,2018]..]

But df.values.tolist() gives

[[2.0,0.18,2018.0],...]

Answer 1

itertuples

list(map(list, df.itertuples(index=False)))

[[2, 0.18, 2018],
 [3, 0.0, 2018],
 [4, 12.48, 2018],
 [6, 0.0, 2018],
 [7, 1.89, 2018]]

---

A little more terse

And far less readable

[*map(list, zip(*map(df.get, df)))]

[[2, 0.18, 2018],
 [3, 0.0, 2018],
 [4, 12.48, 2018],
 [6, 0.0, 2018],
 [7, 1.89, 2018]]

Answer 2

You can use the intermediate numpy records array to conserve datatypes, and then if you must, convert to a list.

This approach, while being quite fast, will leave you with a list of tuples, as opposed to a list of lists.

---

df.to_records(index=False).tolist()

[(2, 0.18, 2018),
 (3, 0.0, 2018),
 (4, 12.48, 2018),
 (6, 0.0, 2018),
 (7, 1.89, 2018)]