input_master.py refactor proposal using GMTK interface #144
Conversation
|
Could you first the linting/build errors reported by the github checks (found by PyFlakes and running |
segway/input_master.py
Outdated
| """ | ||
| returns: iterable of strs containing GMTK parameter objects starting | ||
| with names | ||
| Return list of head track names. |
There was a problem hiding this comment.
Returns the first track name in each track group
segway/input_master.py
Outdated
| type_name = "DT" | ||
| copy_attrs = ParamSpec.copy_attrs + ["seg_countdowns_initial", | ||
| "supervision_type"] | ||
| def generate_gmtk_obj_names(self, obj, track_names): |
segway/input_master.py
Outdated
| type_name = "DT" | ||
| copy_attrs = ParamSpec.copy_attrs + ["seg_countdowns_initial", | ||
| "supervision_type"] | ||
| def generate_gmtk_obj_names(self, obj, track_names): |
There was a problem hiding this comment.
Why is track_names used here? There's self.track_names and even self.track_groups? Is this called multiple times across track groups?
segway/input_master.py
Outdated
| MX: "mx" | ||
| MC: "mc_diag", "mc_gamma", "mc_missing", "gammascale" | ||
| DPMF: "dpmf" | ||
| :param obj: str: type of gmtk object for which names must be generated |
There was a problem hiding this comment.
Maybe instead of obj it should be renamed to gmtk_object_type
segway/input_master.py
Outdated
| """ | ||
| Generate GMTK object names for the types: | ||
| NameCollection: "col" | ||
| entries in NameCollection: "mx_name" |
There was a problem hiding this comment.
It's not actually "mx_name" though right? can you describe this better somehow?
There was a problem hiding this comment.
This has been renamed to "collection_entries" now.
segway/input_master.py
Outdated
| dpmf_names *= multiple # replicate dpmf names for iteration | ||
|
|
||
| # create MX objects | ||
| for i in range(len(names)): |
There was a problem hiding this comment.
use zip(names, dpmf_names, mc_names) to iterate over instead
segway/input_master.py
Outdated
| dpmf_values = str(round(1.0 / self.num_mix_components, | ||
| ROUND_NDIGITS)) | ||
| # create dpmf objects | ||
| for i in range(len(names)): |
There was a problem hiding this comment.
zip(names, values) and for all other instances you can find as well
| @@ -348,6 +401,10 @@ def calc_prob_transition(self, length): | |||
| def make_dense_cpt_segCountDown_seg_segTransition(self): # noqa | |||
| # first values are the ones where segCountDown = 0 therefore | |||
| # the transitions to segTransition = 2 occur early on | |||
|
|
|||
| # see Segway paper | |||
There was a problem hiding this comment.
put the citation in comments (in brackets perhaps)
segway/input_master.py
Outdated
| "covar", "collection_names", "collection_entries", "dpmf", "gammascale", | ||
| "gammashape", "tied_covar"] | ||
|
|
||
| if not gmtk_object_type in allowed_types: |
There was a problem hiding this comment.
allowed_types is not necessary here. You can check if the key exists in GMTK_OBJECT_NAME_GENERATORS
segway/input_master.py
Outdated
| head_track_names = self.get_head_track_names() | ||
| names = [] | ||
| for name in head_track_names: | ||
| names.append("collection_seg_{}".format(name)) |
There was a problem hiding this comment.
For all the GMTK object name format strings, you should define them as constants above. e.g. NAME_COLLECTION_FORMAT_STRING = "collection_seg_{}"
|
This work is continued in PR #163 |
Updated NameCollection, Mean, MC, MX, DPMF, Covar types. In progress: DenseCPT, VE, RealMat