"""
sentiment_app.py
================
Project: Sentiment Analysis Web App
Difficulty: Intermediate
Skills: Python, Flask, scikit-learn, HTML, CSS
Time: High (a week or more)
What you will build:
Build a web application that predicts whether the text
entered by a user has positive, negative, or neutral
sentiment. The project teaches NLP preprocessing, tokenization
and machine learning basics with Flask integration.
How to run:
pip install flask scikit-learn pandas
python sentiment_app.py
Open the local Flask server in your browser and enter sample text to test sentiment predictions.
A sample dataset such as Sentiment140 or IMDb reviews can be used for training and testing the model.
Learning goals:
- Preprocessing text data for NLP tasks
- Converting text into numerical features
- Building a simple sentiment prediction model
- Building a Flask web app and displaying sentiment predictions
Roadmap:
Step 1: Load and inspect the sentiment dataset
Step 2: Clean and preprocess text input data
Step 3: Convert text into numerical feature vectors
Step 4: Build a simple sentiment prediction model
Step 5: Evaluate model accuracy on test data
Step 6: Create Flask routes and an HTML template for structure and CSS for design
Step 7: Connect the trained model to the web app
Step 8: Test predictions with varied user entered text
"""
from flask import Flask, request
from sklearn.feature_extraction.text import CountVectorizer
from sklearn.linear_model import LogisticRegression
# -------------------------------------------------------------------
# Flask App Setup
# -------------------------------------------------------------------
app = Flask(__name__)
# -------------------------------------------------------------------
# Sample Dataset
# -------------------------------------------------------------------
texts = [
"I love this movie",
"This is amazing",
"Fantastic experience",
"I hate this product",
"This is terrible",
"Very disappointing"
]
labels = [
"Positive",
"Positive",
"Positive",
"Negative",
"Negative",
"Negative"
]
# TODO:
# Add more training examples for better predictions
# Include neutral sentiment examples
# Load dataset from a CSV file instead of hardcoding values
# -------------------------------------------------------------------
# Model Training
# -------------------------------------------------------------------
vectorizer = CountVectorizer()
X = vectorizer.fit_transform(texts)
model = LogisticRegression()
model.fit(X, labels)
# -------------------------------------------------------------------
# Core Functions
# -------------------------------------------------------------------
def predict_sentiment(text):
"""
Predict sentiment for user-entered text.
"""
# TODO:
# Add text preprocessing such as:
# - lowercasing
# - punctuation removal
# - stopword filtering
transformed_text = vectorizer.transform([text])
prediction = model.predict(transformed_text)
return prediction[0]
# -------------------------------------------------------------------
# Flask Routes
# -------------------------------------------------------------------
@app.route("/", methods=["GET", "POST"])
def home():
"""
Main route for sentiment prediction.
"""
prediction = ""
if request.method == "POST":
user_text = request.form.get("text")
if user_text:
prediction = predict_sentiment(user_text)
# TODO:
# Move HTML into separate template files
# Add CSS styling
return f"""
Sentiment Analysis App
Sentiment Analysis Web App
{prediction}
"""
# -------------------------------------------------------------------
# Main Program
# -------------------------------------------------------------------
if __name__ == "__main__":
print("Starting Flask server")
app.run(debug=True)