Coverage for dariah_topics/visualization.py : 44%

# -*- coding: utf-8 -*- 

"""LDA visualization. 

This module contains functions for LDA visualization provided by `DARIAH-DE`_. 

.. _Gensim: 

https://radimrehurek.com/gensim/index.html 

.. _Mallet: 

http://mallet.cs.umass.edu/ 

.. _DARIAH-DE: 

https://de.dariah.eu 

https://github.com/DARIAH-DE 

""" 

__author__ = "DARIAH-DE" 

__authors__ = "Steffen Pielstroem, Sina Bock, Severin Simmler" 

__email__ = "pielstroem@biozentrum.uni-wuerzburg.de" 

__version__ = "0.1" 

__date__ = "2017-01-20" 

import logging 

import matplotlib 

matplotlib.use('Agg') 

import matplotlib.pyplot as plt 

import numpy as np 

import os 

import pandas as pd 

import regex 

from collections import defaultdict 

log = logging.getLogger('visualization') 

log.addHandler(logging.NullHandler()) 

logging.basicConfig(level = logging.ERROR, 

format = '%(levelname)s %(name)s: %(message)s') 

def create_doc_topic(corpus, model, doc_labels): 

# Adapted from code by Stefan Pernes 

"""Creates a document-topic data frame. 

Args: 

Gensim corpus. 

Gensim model object. 

List of document labels. 

Returns: Doc_topic DataFrame 

ToDo: 

""" 

no_of_topics = model.num_topics 

no_of_docs = len(doc_labels) 

doc_topic = np.zeros((no_of_topics, no_of_docs)) 

for doc, i in zip(corpus, range(no_of_docs)): # use document bow from corpus 

topic_dist = model.__getitem__(doc) # to get topic distribution from model 

for topic in topic_dist: # topic_dist is a list of tuples 

doc_topic[topic[0]][i] = topic[1] # save topic probability 

topic_labels = [] 

for i in range(no_of_topics): 

topic_terms = [x[0] for x in model.show_topic(i, topn=3)] # show_topic() returns tuples (word_prob, word) 

topic_labels.append(" ".join(topic_terms)) 

doc_topic = pd.DataFrame(doc_topic, index = topic_labels, columns = doc_labels) 

return doc_topic 

def doc_topic_heatmap(data_frame): 

# Adapted from code by Stefan Pernes and Allen Riddell 

"""Plot documnet-topic distribution in a heat map. 

Args: 

Document-topic data frame. 

Returns: Plot with Heatmap 

ToDo: 

""" 

data_frame = data_frame.sort_index() 

doc_labels = list(data_frame.index) 

topic_labels = list(data_frame) 

if len(doc_labels) > 20 or len(topic_labels) > 20: plt.figure(figsize=(20,20)) # if many items, enlarge figure 

plt.pcolor(data_frame, norm=None, cmap='Reds') 

plt.yticks(np.arange(data_frame.shape[0])+1.0, doc_labels) 

plt.xticks(np.arange(data_frame.shape[1])+0.5, topic_labels, rotation='90') 

plt.gca().invert_yaxis() 

plt.tight_layout() 

#plt.savefig(path+"/"+corpusname+"_heatmap.png") #, dpi=80) 

return plt 

def plot_doc_topics(doc_topic, document_index): 

"""Plot topic disctribution in a document. 

Args: 

Document-topic data frame. 

Index of the document to be shown. 

Returns: 

""" 

data = doc_topic[list(doc_topic)[document_index]].copy() 

data = data[data != 0] 

data = data.sort_values() 

values = list(data) 

labels = list(data.index) 

plt.barh(range(len(values)), values, align = 'center', alpha=0.5) 

plt.yticks(range(len(values)), labels) 

plt.title(list(doc_topic)[document_index]) 

plt.xlabel('Proportion') 

plt.ylabel('Topic') 

plt.tight_layout() 

return plt 

try: 

from wordcloud import WordCloud 

# 

# Work in progress following 

# 

def topicwords_in_df(model): 

pattern = regex.compile(r'\p{L}+\p{P}?\p{L}+') 

topics = [] 

index = [] 

for n, topic in enumerate(model.show_topics()): 

topics.append(pattern.findall(topic[1])) 

index.append("Topic " + str(n+1)) 

df = pd.DataFrame(topics, index=index, columns=["Key " + str(x+1) for x in range(len(topics))]) 

return df 

def show_wordle_for_topic(model, topic_nr, words): 

"""Plot wordle for a specific topic 

Args: 

model: Gensim LDA model 

topic_nr(int): Choose topic 

words (int): Number of words to show 

Note: Function does use wordcloud package -> https://pypi.python.org/pypi/wordcloud 

pip install wordcloud 

ToDo: Check if this function should be implemented 

""" 

plt.figure() 

plt.imshow(WordCloud().fit_words(dict(model.show_topic(topic_nr, words)))) 

plt.axis("off") 

plt.title("Topic #" + str(topic_nr + 1)) 

return plt 

def get_color_scale(word, font_size, position, orientation, font_path, random_state=None): 

""" Create color scheme for wordle.""" 

return "hsl(245, 58%, 25%)" # Default. Uniform dark blue. 

#return "hsl(0, 00%, %d%%)" % random.randint(80, 100) # Greys for black background. 

#return "hsl(221, 65%%, %d%%)" % random.randint(30, 35) # Dark blues for white background 

def get_topicRank(topic, topicRanksFile): 

#print("getting topic rank.") 

with open(topicRanksFile, "r") as infile: 

topicRanks = pd.read_csv(infile, sep=",", index_col=0) 

rank = int(topicRanks.iloc[topic]["Rank"]) 

return rank 

def read_mallet_word_weights(word_weights_file): 

"""Read Mallet word_weigths file 

Description: 

Reads Mallet word_weigths into pandas DataFrame. 

Args: 

word_weigts_file: Word_weights_file created with Mallet 

Returns: Pandas DataFrame 

Note: 

ToDo: 

""" 

word_scores = pd.read_table(word_weights_file, header=None, sep="\t") 

word_scores = word_scores.sort(columns=[0,2], axis=0, ascending=[True, False]) 

word_scores_grouped = word_scores.groupby(0) 

return word_scores_grouped 

def get_wordlewords(word_scores_grouped, number_of_top_words, topic_nr): 

"""Transform Mallet output for wordle generation. 

Description: 

Get words for wordle. 

Args: 

word_scores_grouped(DataFrame): Uses read_mallet_word_weights() to get 

grouped word scores. 

topic_nr(int): Topic the wordle should be generated for 

number_of_top_words(int): Number of top words that should be considered 

Returns: Words for wordle. 

Note: 

ToDo: 

""" 

topic_word_scores = word_scores_grouped.get_group(topic_nr) 

top_topic_word_scores = topic_word_scores.iloc[0:number_of_top_words] 

topic_words = top_topic_word_scores.loc[:,1].tolist() 

#word_scores = top_topic_word_scores.loc[:,2].tolist() 

wordlewords = "" 

j = 0 

for word in topic_words: 

word = word 

#score = word_scores[j] 

j += 1 

wordlewords = wordlewords + ((word + " ")) 

return wordlewords 

def plot_wordle_from_mallet(word_weights_file, 

topic_nr, 

number_of_top_words, 

outfolder, 

dpi): 

"""Generate wordles from Mallet output. 

Description: 

This function does use the wordcloud module to plot wordles. 

Uses read_mallet_word_weigths() and get_wordlewords() to get 

word_scores and words for wordle. 

Args: 

word_weigts_file: Word_weights_file created with Mallet 

topic_nr(int): Topic the wordle should be generated for 

number_of_top_words(int): Number of top words that should be considered 

for the wordle 

outfolder(str): Specify path to safe wordle. 

dpi(int): Set resolution for wordle. 

Returns: Plot 

Note: 

ToDo: 

""" 

word_scores_grouped = read_mallet_word_weights(word_weights_file) 

text = get_wordlewords(word_scores_grouped, number_of_top_words, topic_nr) 

wordcloud = WordCloud(width=600, height=400, background_color="white", margin=4).generate(text) 

default_colors = wordcloud.to_array() 

figure_title = "topic "+ str(topic_nr) 

plt.imshow(default_colors) 

plt.imshow(wordcloud) 

plt.title(figure_title, fontsize=30) 

plt.axis("off") 

## Saving the image file. 

if not os.path.exists(outfolder): 

os.makedirs(outfolder) 

figure_filename = "wordle_tp"+"{:03d}".format(topic_nr) + ".png" 

plt.savefig(outfolder + figure_filename, dpi=dpi) 

return plt 

def plot_wordle_from_lda(model, vocab, topic_nr, words, height, width): 

topic_dist = model.topic_word_[topic_nr] 

topic_words = np.array(vocab)[np.argsort(topic_dist)][:-words:-1] 

token_value = {} 

for token, value in zip(topic_words, topic_dist[:-words:-1]): 

token_value.update({token: value}) 

return WordCloud(background_color='white', height=height, width=width).fit_words(token_value) 

except ImportError as e: 

log.info('WordCloud functions not available, they require the wordcloud module') 

def doc_topic_heatmap_interactive(doc_topic, title): 

"""Plot interactive doc_topic_heatmap 

Description: 

With this function you can plot an interactive doc_topic matrix. 

Args: 

doc_topic: Doc_topic matrix in a DataFrame 

title(str): Title shown in the plot. 

Returns: bokeh plot 

Note: 

ToDo: 

""" 

log.info("Importing functions from bokeh ...") 

try: 

#from ipywidgets import interact 

from bokeh.io import output_notebook 

from bokeh.plotting import figure 

from math import pi 

from bokeh.models import ( 

ColumnDataSource, 

HoverTool, 

LinearColorMapper, 

BasicTicker, 

ColorBar 

) 

output_notebook() 

documents = list(doc_topic.columns) 

topics = doc_topic.index 

score = [] 

for x in doc_topic.apply(tuple): 

score.extend(x) 

data = { 

'Topic': list(doc_topic.index) * len(doc_topic.columns), 

'Document': [item for item in list(doc_topic.columns) for i in range(len(doc_topic.index))], 

'Score': score 

} 

df = doc_topic.from_dict(data) 

colors = ["#c6dbef", "#9ecae1", "#6baed6", "#4292c6", "#2171b5", "#08519c", "#08306b"] 

mapper = LinearColorMapper(palette=colors, low=df.Score.min(), high=df.Score.max()) 

source = ColumnDataSource(df) 

TOOLS = "hover,save,pan,box_zoom,reset,wheel_zoom" 

p = figure(title=title, 

x_range=documents, y_range=list(reversed(topics)), 

x_axis_location="above", plot_width=1024, plot_height=768, 

tools=TOOLS, toolbar_location='below', responsive=True) 

p.grid.grid_line_color = None 

p.axis.axis_line_color = None 

p.axis.major_tick_line_color = None 

p.axis.major_label_text_font_size = "9pt" 

p.axis.major_label_standoff = 0 

p.xaxis.major_label_orientation = pi / 3 

p.rect(x="Document", y="Topic", width=1, height=1, 

source=source, 

fill_color={'field': 'Score', 'transform': mapper}, 

line_color=None) 

color_bar = ColorBar(color_mapper=mapper, major_label_text_font_size="10pt", 

ticker=BasicTicker(desired_num_ticks=len(colors)), 

label_standoff=6, border_line_color=None, location=(0, 0)) 

p.add_layout(color_bar, 'right') 

p.select_one(HoverTool).tooltips = [ 

('Document', '@Document'), 

('Topic', '@Topic'), 

('Score', '@Score') 

] 

return p 

except: 

log.info("Bokeh could not be imported now using mathplotlib") 

doc_topic_heatmap(doc_topic) 

p.add_layout(color_bar, 'right') 

p.select_one(HoverTool).tooltips = [ 

('Document', '@Document'), 

('Topic', '@Topic'), 

('Score', '@Score') 

] 

return p 

def show_topic_over_time(doc_topic, labels=['armee truppen general', 'regierung preußen partei', 'dichter goethe kunst'], threshold=0.1, starttime=1841, endtime=1920): 

"""Creates a visualization that shows topics over time 

Description: 

With this function you can plot topics over time using metadata stored in the documents name. 

Only works with mallet output. 

Args: 

doc_topic: doc-topic matrix created by mallet.show_doc_topic_matrix 

labels(list[str]): first three keys in a topic to select 

threshold(float): threshold set to define if a topic in a document is viable 

starttime(int): sets starting point for visualization 

endtime(int): sets ending point for visualization 

Returns: matplotlib plot 

Note: this function is created for a corpus with filenames that looks like: 

1866_ArticleName.txt 

ToDo: make it compatible with gensim output 

""" 

years=list(range(starttime,endtime)) 

doc_topicT = doc_topic.T 

for label in labels: 

topic_over_threshold_per_year =[] 

df = doc_topicT.loc[doc_topicT[label] > threshold] 

d = defaultdict(int) 

for item in df.index.values: 

year = item.split('_') 

d[year[0]]+=1 

for year in years: 

topic_over_threshold_per_year.append(d[str(year)]) 

plt.plot(years, topic_over_threshold_per_year, label=label) 

plt.xlabel('Year') 

plt.ylabel('count topics over threshold') 

plt.legend() 

fig = plt.gcf() 

fig.set_size_inches(18.5, 10.5) 

plt.show()