Components by Term Frame

Example

>>> from sklearn.decomposition import LatentDirichletAllocation
>>> lda = LatentDirichletAllocation(
...     n_components=10,
...     learning_decay=0.7,
...     learning_offset=50.0,
...     max_iter=10,
...     batch_size=128,
...     evaluate_every=-1,
...     perp_tol=0.1,
...     mean_change_tol=0.001,
...     max_doc_update_iter=100,
...     random_state=0,
... )
>>> from techminer2.packages.topic_modeling.user import ComponentsByTermDataFrame
>>> df = (
...     ComponentsByTermDataFrame()
...     #
...     # FIELD:
...     .with_field("raw_descriptors")
...     .having_terms_in_top(50)
...     .having_terms_ordered_by("OCC")
...     .having_term_occurrences_between(None, None)
...     .having_term_citations_between(None, None)
...     .having_terms_in(None)
...     #
...     # DECOMPOSITION:
...     .using_decomposition_algorithm(lda)
...     .using_top_terms_by_theme(5)
...     #
...     # TFIDF:
...     .using_binary_term_frequencies(False)
...     .using_row_normalization(None)
...     .using_idf_reweighting(False)
...     .using_idf_weights_smoothing(False)
...     .using_sublinear_tf_scaling(False)
...     #
...     # DATABASE:
...     .where_root_directory_is("examples/fintech/")
...     .where_database_is("main")
...     .where_record_years_range_is(None, None)
...     .where_record_citations_range_is(None, None)
...     .where_records_match(None)
...     #
...     .run()
... )
>>> df  
term       FINTECH 38:6131  ...  REGULATION 03:0461
component                   ...
0                 8.362389  ...            2.099995
1                 3.334489  ...            0.100000
2                 2.099949  ...            0.100000
3                 0.100000  ...            0.100000
4                 0.100000  ...            0.100000
5                 4.837552  ...            0.100000
6                 4.779077  ...            0.100000
7                 3.100090  ...            0.100000
8                 1.186569  ...            1.100005
9                11.099884  ...            0.100000

[10 rows x 50 columns]