| 345 | ], |
| 346 | ) |
| 347 | def test_describe_numeric(method, test_values): |
| 348 | # prepare test case which approx quantiles will be the same as actuals |
| 349 | s = pd.Series(list(range(test_values[1])) * test_values[0]) |
| 350 | df = pd.DataFrame( |
| 351 | { |
| 352 | "a": list(range(test_values[1])) * test_values[0], |
| 353 | "b": list(range(test_values[0])) * test_values[1], |
| 354 | } |
| 355 | ) |
| 356 | |
| 357 | ds = dd.from_pandas(s, test_values[0]) |
| 358 | ddf = dd.from_pandas(df, test_values[0]) |
| 359 | |
| 360 | test_quantiles = [0.25, 0.75] |
| 361 | |
| 362 | assert_eq(df.describe(), ddf.describe(percentiles_method=method)) |
| 363 | assert_eq(s.describe(), ds.describe(percentiles_method=method)) |
| 364 | |
| 365 | assert_eq( |
| 366 | df.describe(percentiles=test_quantiles), |
| 367 | ddf.describe(percentiles=test_quantiles, percentiles_method=method), |
| 368 | ) |
| 369 | assert_eq(s.describe(), ds.describe(split_every=2, percentiles_method=method)) |
| 370 | assert_eq(df.describe(), ddf.describe(split_every=2, percentiles_method=method)) |
| 371 | |
| 372 | # remove string columns |
| 373 | df = pd.DataFrame( |
| 374 | { |
| 375 | "a": list(range(test_values[1])) * test_values[0], |
| 376 | "b": list(range(test_values[0])) * test_values[1], |
| 377 | "c": list("abcdef"[: test_values[0]]) * test_values[1], |
| 378 | } |
| 379 | ) |
| 380 | ddf = dd.from_pandas(df, test_values[0]) |
| 381 | assert_eq(df.describe(), ddf.describe(percentiles_method=method)) |
| 382 | assert_eq(df.describe(), ddf.describe(split_every=2, percentiles_method=method)) |
| 383 | |
| 384 | |
| 385 | @pytest.mark.parametrize( |