Visualizing Sentiment Analysis Results in Python using Matplotlib

In this article, we will add graphical representation of the sentiment analysis results using Matplotlib. The goal is to visualize the sentiment scores of multiple sentences, with a bar chart that distinguishes positive and negative sentiments using different colors.

Pre-requisites

Make sure you have the following libraries installed:

pip install transformers torch matplotlib

• transformers: For handling pre-trained NLP models.
• torch: For running the model.
• matplotlib: For creating the graphical representation of sentiment analysis results.

Python Code with Visualization

Here’s the Python code that integrates sentiment analysis with data visualization.

import matplotlib.pyplot as plt
from transformers import pipeline
from transformers import AutoTokenizer, AutoModelForSequenceClassification

# Load pre-trained model and tokenizer
model_name = "distilbert-base-uncased-finetuned-sst-2-english"
model = AutoModelForSequenceClassification.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)

# Initialize the sentiment-analysis pipeline
classifier = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer)

# List of 10 sentences for sentiment analysis
sentences = [
    "I love you! I love you! I love you!",
    "I feel so sad today.",
    "This is the best day ever!",
    "I can't stand the rain.",
    "Everything is going so well.",
    "I hate waiting in line.",
    "The weather is nice, but it's cold.",
    "I'm so proud of my achievements.",
    "I am very upset with the decision.",
    "I am feeling optimistic about the future."
]

# Prepare data for the chart
scores = []
colors = []

for sentence in sentences:
    result = classifier(sentence)
    sentiment = result[0]['label']
    score = result[0]['score']
    scores.append(score)

    # Color bars based on sentiment: Positive -> green, Negative -> red
    if sentiment == "POSITIVE":
        colors.append("green")
    else:
        colors.append("red")

# Create a bar chart
plt.figure(figsize=(10, 6))
bars = plt.bar(sentences, scores, color=colors)

# Add labels and title with a line break
plt.xlabel('Sentences')
plt.ylabel('Sentiment Score')
plt.title('Sentiment Analysis of 10 Sentences\n')  # Added newline here
plt.xticks(rotation=45, ha="right")

# Adjust spacing with top margin (to add ceiling space)
plt.subplots_adjust(top=0.85)  # Adjust the top spacing (20px roughly equivalent to 0.1 top margin)

plt.tight_layout()  # Adjusts the rest of the layout

# Display the sentiment score on top of the bars
for bar in bars:
    yval = bar.get_height()
    plt.text(bar.get_x() + bar.get_width() / 2, yval + 0.02, f'{yval:.2f}', ha='center', va='bottom', fontsize=9)

# Show the plot
plt.show()

Breakdown of Code

Importing Necessary Libraries:
We import matplotlib.pyplot to create plots and transformers to perform sentiment analysis.

   import matplotlib.pyplot as plt
   from transformers import pipeline
   from transformers import AutoTokenizer, AutoModelForSequenceClassification

Loading Pre-trained Model:
We load the DistilBERT model fine-tuned for sentiment analysis on the SST-2 dataset. We also load the associated tokenizer that converts text into model-readable tokens.

   model_name = "distilbert-base-uncased-finetuned-sst-2-english"
   model = AutoModelForSequenceClassification.from_pretrained(model_name)
   tokenizer = AutoTokenizer.from_pretrained(model_name)

Initializing Sentiment Analysis Pipeline:
The classifier pipeline is set up for sentiment analysis. This pipeline takes care of tokenizing the input text, performing inference, and returning the result.

   classifier = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer)

Sentences for Sentiment Analysis:
We create a list of 10 sentences to analyze. Each sentence is a unique expression of sentiment, ranging from very positive to negative.

   sentences = [
       "I love you! I love you! I love you!",
       "I feel so sad today.",
       "This is the best day ever!",
       "I can't stand the rain.",
       "Everything is going so well.",
       "I hate waiting in line.",
       "The weather is nice, but it's cold.",
       "I'm so proud of my achievements.",
       "I am very upset with the decision.",
       "I am feeling optimistic about the future."
   ]

Processing Sentiment and Preparing Data:
For each sentence, we classify its sentiment and extract the score. Based on the sentiment label (POSITIVE or NEGATIVE), we assign a color for the bars in the chart. Positive sentences will be green, while negative ones will be red.

   scores = []
   colors = []

   for sentence in sentences:
       result = classifier(sentence)
       sentiment = result[0]['label']
       score = result[0]['score']
       scores.append(score)

       if sentiment == "POSITIVE":
           colors.append("green")
       else:
           colors.append("red")

Creating the Bar Chart:
We use matplotlib to create a bar chart. The height of each bar represents the sentiment score for a sentence, and the colors differentiate the positive and negative sentiments.

   plt.figure(figsize=(10, 6))
   bars = plt.bar(sentences, scores, color=colors)

Adding Labels and Adjusting Layout:
We customize the appearance of the plot by rotating the x-axis labels for better readability, adding a title, and adjusting the layout for optimal spacing.

   plt.xlabel('Sentences')
   plt.ylabel('Sentiment Score')
   plt.title('Sentiment Analysis of 10 Sentences\n')  # Added newline here
   plt.xticks(rotation=45, ha="right")
   plt.subplots_adjust(top=0.85)  # Adjust the top spacing
   plt.tight_layout()  # Adjusts the rest of the layout

Displaying Sentiment Scores on Top of Bars:
We also display the sentiment score on top of each bar to make the chart more informative.

   for bar in bars:
       yval = bar.get_height()
       plt.text(bar.get_x() + bar.get_width() / 2, yval + 0.02, f'{yval:.2f}', ha='center', va='bottom', fontsize=9)

Displaying the Plot:
Finally, the chart is displayed using plt.show(), which renders the plot.

   plt.show()

Sample Output

The output of this code will be a bar chart displaying the sentiment scores of the 10 sentences. Positive sentences will be represented by green bars, while negative sentences will be shown as red bars.

Conclusion

By combining sentiment analysis with data visualization, we can better interpret the emotional tone behind textual data. The graphical representation in this article offers a clearer understanding of the sentiment distribution, allowing you to easily spot trends in the text. You can apply this technique to various use cases like analyzing product reviews, social media posts, or customer feedback.

With the powerful combination of Hugging Face's transformers and matplotlib, this workflow can be extended and customized to suit various NLP tasks.