def test_mat2d_window_from_indices(mat, n_row_indices, n_col_indices, copy):
mat = np.array(mat)
n_rows, n_cols = mat.shape
if n_row_indices == 0:
row_indices = None
else:
row_indices = np.random.choice(np.arange(n_rows),
size=min(n_rows, n_row_indices),
replace=False)
if n_col_indices == 0:
col_indices = None
else:
col_indices = np.random.choice(np.arange(n_cols),
size=min(n_cols, n_col_indices),
replace=False)
window = mat2d_window_from_indices(mat, row_indices, col_indices, copy)
if row_indices is None:
asserted_y_shape = n_rows
else:
asserted_y_shape = len(row_indices)
assert window.shape[0] == asserted_y_shape
if col_indices is None:
asserted_x_shape = n_cols
else:
asserted_x_shape = len(col_indices)
assert window.shape[1] == asserted_x_shape
if row_indices is None:
row_indices_check = np.arange(n_rows)
else:
row_indices_check = row_indices
if col_indices is None:
col_indices_check = np.arange(n_cols)
else:
col_indices_check = col_indices
for w_y, m_y in enumerate(row_indices_check):
for w_x, m_x in enumerate(col_indices_check):
assert window[w_y, w_x] == mat[m_y, m_x]
def plot_topic_word_heatmap(fig,
ax,
topic_word_distrib,
vocab,
which_topics=None,
which_topic_indices=None,
which_words=None,
which_word_indices=None,
xaxislabel=None,
yaxislabel=None,
**kwargs):
"""
Plot a heatmap for a topic-word distribution `topic_word_distrib` to a matplotlib Figure `fig` and Axes `ax`
using `vocab` as vocabulary on the x-axis and topics from 1 to `n_topics=doc_topic_distrib.shape[1]` on
the y-axis.
A subset of words from `vocab` can be specified either directly with a sequence `which_words` or
`which_document_indices` containing a sequence of word indices in `vocab`.
A subset of topics can be specified either with a sequence `which_topics` containing sequence of numbers between
[1, n_topics] or `which_topic_indices` which is a number between [0, n_topics-1]
Additional arguments can be passed via `kwargs` to `plot_heatmap`.
Please note that it is almost always necessary to select a subset of your topic-word distribution with the
`which_words` or `which_topics` parameters, as otherwise the amount of data to be plotted will be too high
to give a reasonable picture.
"""
if which_topics is not None and which_topic_indices is not None:
raise ValueError(
'only `which_topics` or `which_topic_indices` can be set, not both'
)
if which_words is not None and which_word_indices is not None:
raise ValueError(
'only `which_words` or `which_word_indices` can be set, not both')
if which_topics is not None:
which_topic_indices = np.array(which_topics) - 1
if which_words is not None:
which_word_indices = np.where(np.isin(vocab, which_words))[0]
select_distrib_subset = False
topic_labels = np.array(range(1, topic_word_distrib.shape[0] + 1))
if which_topic_indices is not None:
select_distrib_subset = True
topic_labels = topic_labels[which_topic_indices]
if which_word_indices is not None:
select_distrib_subset = True
vocab = np.array(vocab)[which_word_indices]
if select_distrib_subset:
topic_word_distrib = mat2d_window_from_indices(topic_word_distrib,
which_topic_indices,
which_word_indices)
return plot_heatmap(fig,
ax,
topic_word_distrib,
xaxislabel=xaxislabel or 'vocab',
yaxislabel=yaxislabel or 'topic',
xticklabels=vocab,
yticklabels=topic_labels,
**kwargs)
def plot_doc_topic_heatmap(fig,
ax,
doc_topic_distrib,
doc_labels,
topic_labels=None,
which_documents=None,
which_document_indices=None,
which_topics=None,
which_topic_indices=None,
xaxislabel=None,
yaxislabel=None,
**kwargs):
"""
Plot a heatmap for a document-topic distribution `doc_topic_distrib` to a matplotlib Figure `fig` and Axes `ax`
using `doc_labels` as document labels on the y-axis and topics from 1 to `n_topics=doc_topic_distrib.shape[1]` on
the x-axis.
Custom topic labels can be passed as `topic_labels`.
A subset of documents can be specified either with a sequence `which_documents` containing a subset of document
labels from `doc_labels` or `which_document_indices` containing a sequence of document indices.
A subset of topics can be specified either with a sequence `which_topics` containing sequence of numbers between
[1, n_topics] or `which_topic_indices` which is a number between [0, n_topics-1]
Additional arguments can be passed via `kwargs` to `plot_heatmap`.
Please note that it is almost always necessary to select a subset of your document-topic distribution with the
`which_documents` or `which_topics` parameters, as otherwise the amount of data to be plotted will be too high
to give a reasonable picture.
"""
if which_documents is not None and which_document_indices is not None:
raise ValueError(
'only `which_documents` or `which_document_indices` can be set, not both'
)
if which_topics is not None and which_topic_indices is not None:
raise ValueError(
'only `which_topics` or `which_topic_indices` can be set, not both'
)
if which_documents is not None:
which_document_indices = np.where(np.isin(doc_labels,
which_documents))[0]
if which_topics is not None:
which_topic_indices = np.array(which_topics) - 1
select_distrib_subset = False
if topic_labels is None:
topic_labels = np.array(range(1, doc_topic_distrib.shape[1] + 1))
elif not isinstance(topic_labels, np.ndarray):
topic_labels = np.array(topic_labels)
if which_document_indices is not None:
select_distrib_subset = True
doc_labels = np.array(doc_labels)[which_document_indices]
if which_topic_indices is not None:
select_distrib_subset = True
topic_labels = topic_labels[which_topic_indices]
if select_distrib_subset:
doc_topic_distrib = mat2d_window_from_indices(doc_topic_distrib,
which_document_indices,
which_topic_indices)
return plot_heatmap(fig,
ax,
doc_topic_distrib,
xaxislabel=xaxislabel or 'topic',
yaxislabel=yaxislabel or 'document',
xticklabels=topic_labels,
yticklabels=doc_labels,
**kwargs)
def plot_topic_word_heatmap(fig,
ax,
topic_word_distrib,
vocab,
topic_labels=None,
which_topics=None,
which_topic_indices=None,
which_words=None,
which_word_indices=None,
xaxislabel=None,
yaxislabel=None,
**kwargs):
"""
Plot a heatmap for a topic-word distribution `topic_word_distrib` to a matplotlib Figure `fig` and Axes `ax`
using `vocab` as vocabulary on the x-axis and topics from 1 to `n_topics=doc_topic_distrib.shape[1]` on
the y-axis.
.. note:: It is almost always necessary to select a subset of your topic-word distribution with the
`which_words` or `which_topics` parameters, as otherwise the amount of data to be plotted will be too high
to give a reasonable picture.
:param fig: matplotlib Figure object
:param ax: matplotlib Axes object
:param topic_word_distrib: topic-word distribution; shape KxM, where K is number of topics, M is vocabulary size
:param vocab: vocabulary array of length M
:param topic_labels: labels used for each row; either single format string with
placeholders ``"{i0}"`` (zero-based topic index) or ``"{i1}"`` (one-based topic index), or
list of topic label strings
:param which_topics: select topics via topic label strings (when string array or list and `topic_labels` is given)
or with one-based topic index in [1, K] (when integer array or list)
:param which_topic_indices: alternatively, select topics with zero-based topic index in [0, K-1]
:param which_words: select words with one-based word index in [1, M]
:param which_word_indices: alternatively, select words with zero-based word index in [0, K-1]
:param xaxislabel: x axis label string
:param yaxislabel: y axis label string
:param kwargs: additional arguments passed to :func:`~tmtoolkit.topicmod.visualize.plot_heatmap`
:return: tuple of generated (matplotlib Figure object, matplotlib Axes object)
"""
if not isinstance(topic_word_distrib,
np.ndarray) or topic_word_distrib.ndim != 2:
raise ValueError('`mat` must be a 2D NumPy array')
if topic_word_distrib.shape[0] == 0 or topic_word_distrib.shape[1] == 0:
raise ValueError('invalid shape for `mat`: %s' %
str(topic_word_distrib.shape))
if which_topics is not None and which_topic_indices is not None:
raise ValueError(
'only `which_topics` or `which_topic_indices` can be set, not both'
)
if which_words is not None and which_word_indices is not None:
raise ValueError(
'only `which_words` or `which_word_indices` can be set, not both')
if which_words is not None:
which_word_indices = np.where(np.isin(vocab, which_words))[0]
select_distrib_subset = False
if topic_labels is None:
topic_labels = np.array(range(1, topic_word_distrib.shape[0] + 1))
elif not isinstance(topic_labels, np.ndarray):
topic_labels = np.array(topic_labels)
if which_topics is not None:
which_topics = np.array(which_topics)
if np.issubdtype(which_topics.dtype, np.str):
which_topic_indices = np.where(np.isin(topic_labels,
which_topics))[0]
else:
which_topic_indices = which_topics - 1
if which_topic_indices is not None:
select_distrib_subset = True
topic_labels = topic_labels[which_topic_indices]
if which_word_indices is not None:
select_distrib_subset = True
vocab = np.array(vocab)[which_word_indices]
if select_distrib_subset:
topic_word_distrib = mat2d_window_from_indices(topic_word_distrib,
which_topic_indices,
which_word_indices)
return plot_heatmap(fig,
ax,
topic_word_distrib,
xaxislabel=xaxislabel or 'vocab',
yaxislabel=yaxislabel or 'topic',
xticklabels=vocab,
yticklabels=topic_labels,
**kwargs)
def plot_doc_topic_heatmap(fig,
ax,
doc_topic_distrib,
doc_labels,
topic_labels=None,
which_documents=None,
which_document_indices=None,
which_topics=None,
which_topic_indices=None,
xaxislabel=None,
yaxislabel=None,
**kwargs):
"""
Plot a heatmap for a document-topic distribution `doc_topic_distrib` to a matplotlib Figure `fig` and Axes `ax`
using `doc_labels` as document labels on the y-axis and topics from 1 to K (number of topics) on
the x-axis.
.. note:: It is almost always necessary to select a subset of your document-topic distribution with the
`which_documents` or `which_topics` parameters, as otherwise the amount of data to be plotted will be too
high to give a reasonable picture.
:param fig: matplotlib Figure object
:param ax: matplotlib Axes object
:param doc_topic_distrib: document-topic distribution; shape NxK, where N is the number of documents, K is the
number of topics
:param doc_labels: list/array of length N with a string label for each document
:param topic_labels: labels used for each row; either single format string with
placeholders ``"{i0}"`` (zero-based topic index) or ``"{i1}"`` (one-based topic index), or
list of topic label strings
:param which_documents: select documents via document label strings
:param which_document_indices: alternatively, select documents with zero-based document index in [0, N-1]
:param which_topics: select topics via topic label strings (when string array or list) or with
one-based topic index in [1, K] (when integer array or list)
:param which_topic_indices: alternatively, select topics with zero-based topic index in [0, K-1]
:param xaxislabel: x axis label string
:param yaxislabel: y axis label string
:param kwargs: additional arguments passed to :func:`~tmtoolkit.topicmod.visualize.plot_heatmap`
:return: tuple of generated (matplotlib Figure object, matplotlib Axes object)
"""
if not isinstance(doc_topic_distrib,
np.ndarray) or doc_topic_distrib.ndim != 2:
raise ValueError('`mat` must be a 2D NumPy array')
if doc_topic_distrib.shape[0] == 0 or doc_topic_distrib.shape[1] == 0:
raise ValueError('invalid shape for `mat`: %s' %
str(doc_topic_distrib.shape))
if which_documents is not None and which_document_indices is not None:
raise ValueError(
'only `which_documents` or `which_document_indices` can be set, not both'
)
if which_topics is not None and which_topic_indices is not None:
raise ValueError(
'only `which_topics` or `which_topic_indices` can be set, not both'
)
if which_documents is not None:
which_document_indices = np.where(np.isin(doc_labels,
which_documents))[0]
select_distrib_subset = False
if topic_labels is None:
topic_labels = np.array(range(1, doc_topic_distrib.shape[1] + 1))
elif not isinstance(topic_labels, np.ndarray):
topic_labels = np.array(topic_labels)
if which_topics is not None:
which_topics = np.array(which_topics)
if which_topics.dtype.kind == 'U':
which_topic_indices = np.where(np.isin(topic_labels,
which_topics))[0]
else:
which_topic_indices = which_topics - 1
if which_document_indices is not None:
select_distrib_subset = True
doc_labels = np.array(doc_labels)[which_document_indices]
if which_topic_indices is not None:
select_distrib_subset = True
topic_labels = topic_labels[which_topic_indices]
if select_distrib_subset:
doc_topic_distrib = mat2d_window_from_indices(doc_topic_distrib,
which_document_indices,
which_topic_indices)
return plot_heatmap(fig,
ax,
doc_topic_distrib,
xaxislabel=xaxislabel or 'topic',
yaxislabel=yaxislabel or 'document',
xticklabels=topic_labels,
yticklabels=doc_labels,
**kwargs)
