COMPANY : CODTECH IT SOLUTIONS

NAME : NAVEEN K

INTERN ID : CT06DA721

DOMAIN : MACHINE LEARNING

DURATION : 6 WEEKS

MENTOR : NEELA SANTHOSH KUMAR

Task 2 – Sentiment Analysis using TF-IDF and Logistic Regression

Internship Project - Machine Learning (CodTech)

Hello! This repository contains the solution to Task 2 of the CodTech Machine Learning Internship. The task was to perform Sentiment Analysis on a dataset of customer reviews using TF-IDF vectorization and a Logistic Regression classifier.

The objective is to automatically determine whether a customer review expresses a positive or negative sentiment.

---

Objective

To build a complete NLP pipeline that:

• Takes a dataset of customer reviews
• Preprocesses the text to make it clean and readable
• Converts the text into numerical form using TF-IDF
• Trains a Logistic Regression model to classify sentiment
• Evaluates and visualizes model performance

---

Dataset

I used a dataset converted from amazon_cells_labelled.txt into a CSV format. It contains 999 reviews with two columns:

• review – The customer’s review text
• sentiment – The corresponding label (1 = positive, 0 = negative)

---

Step-by-Step Workflow

🔹 Step 1: Data Loading The CSV file was loaded into a pandas DataFrame, and column names were stripped of hidden spaces to avoid errors.

🔹 Step 2: Data Exploration I checked for class balance, verified data types, and printed samples to understand the structure.

🔹 Step 3: Text Preprocessing A cleaning function was used to:

• Convert text to lowercase
• Remove punctuation, numbers, and extra spaces

The cleaned text was stored in a new column called cleaned_review.

🔹 Step 4: TF-IDF Vectorization Using TfidfVectorizer, I transformed the cleaned text into a numerical matrix of TF-IDF features. This step converts natural language into a format understandable by the model.

🔹 Step 5: Train-Test Split The dataset was split 80:20 into training and testing sets.

🔹 Step 6: Logistic Regression Model A LogisticRegression model from scikit-learn was trained on the TF-IDF vectors.

🔹 Step 7: Model Evaluation I predicted sentiments for the test set and evaluated performance using:

• Accuracy Score
• Classification Report (Precision, Recall, F1)

Output: