CLN: use self.dtype internally in Categorical

[topper-123]

Some clean-up that makes the code a bit easier to read IMO.

[pep8speaks]

Hello @topper-123! Thanks for updating the PR.

Cheers ! There are no PEP8 issues in this Pull Request. 🍻

Comment last updated on August 26, 2018 at 17:24 Hours UTC

[jreback]

looks fine, can u do a perf check

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (master@fa47b8d). Click here to learn what that means.
The diff coverage is 71.42%.

@@            Coverage Diff            @@
##             master   #22513   +/-   ##
=========================================
  Coverage          ?   92.03%           
=========================================
  Files             ?      169           
  Lines             ?    50780           
  Branches          ?        0           
=========================================
  Hits              ?    46737           
  Misses            ?     4043           
  Partials          ?        0

Flag	Coverage Δ
#multiple	90.44% <71.42%> (?)
#single	42.22% <14.28%> (?)

Impacted Files	Coverage Δ
pandas/core/arrays/categorical.py	95.75% <71.42%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update fa47b8d...70ddcef. Read the comment docs.

[topper-123]

Ran ASV on benchmarks/categoricals.py. Result are varying a bit, which I attribute to randomness. For example:

      before           after         ratio
     [fa47b8d9]       [70ddcef7]
+          5.65μs           13.7μs     2.42  categoricals.CategoricalSlicing.time_getitem_list_like('monotonic_incr')
+          5.76ms           6.70ms     1.16  categoricals.CategoricalSlicing.time_getitem_bool_array('monotonic_incr')
+          3.42μs           3.91μs     1.14  categoricals.CategoricalSlicing.time_getitem_scalar('monotonic_decr')
+           610μs            689μs     1.13  categoricals.CategoricalSlicing.time_getitem_list('monotonic_decr')
-          6.12μs           5.55μs     0.91  categoricals.CategoricalSlicing.time_getitem_list_like('monotonic_decr')
-          6.46μs           5.65μs     0.88  categoricals.CategoricalSlicing.time_getitem_list_like('non_monotonic')
-          3.66μs           3.20μs     0.88  categoricals.CategoricalSlicing.time_getitem_scalar('monotonic_incr')
-          6.08ms           5.21ms     0.86  categoricals.CategoricalSlicing.time_getitem_bool_array('monotonic_decr')
-          4.66μs           4.00μs     0.86  categoricals.CategoricalSlicing.time_getitem_slice('non_monotonic')
-           729μs            567μs     0.78  categoricals.CategoricalSlicing.time_getitem_list('monotonic_incr')

SOME BENCHMARKS HAVE CHANGED SIGNIFICANTLY.

and

      before           after         ratio
     [fa47b8d9]       [70ddcef7]
+          15.6ms           26.0ms     1.67  categoricals.CategoricalSlicing.time_getitem_bool_array('non_monotonic')
+           648μs            933μs     1.44  categoricals.CategoricalSlicing.time_getitem_list('monotonic_decr')
+           613μs            854μs     1.39  categoricals.CategoricalSlicing.time_getitem_list('monotonic_incr')
+          4.88μs           6.10μs     1.25  categoricals.CategoricalSlicing.time_getitem_list_like('monotonic_incr')
-          3.66μs           3.20μs     0.88  categoricals.CategoricalSlicing.time_getitem_scalar('monotonic_incr')
-          4.39μs           3.50μs     0.80  categoricals.CategoricalSlicing.time_getitem_scalar('non_monotonic')
-           729μs            539μs     0.74  categoricals.CategoricalSlicing.time_getitem_list('non_monotonic')
-          6.70ms           4.69ms     0.70  categoricals.CategoricalSlicing.time_getitem_bool_array('monotonic_incr')

SOME BENCHMARKS HAVE CHANGED SIGNIFICANTLY.

Notice how categoricals.CategoricalSlicing.time_getitem_bool_array('monotonic_incr') in one run is slow (1.16) and th other fast (0.70).

Overall, I'd say this has is no significant effect on speed.

[gfyoung]

The benchmarks look alright to me.

cc @jreback

[jreback]

@TomAugspurger a look pls. lgtm to me.