Classification using TF-IDF

In this subsection, we will perform sentiment analysis and text classification based on TF-IDF.  Here, TF-IDF is generated using the scikit-learn library.  Let's see how we perform sentiment analysis using TF-IDF features using the following steps:

1. Load the dataset:

The first step for building a machine learning model is to load the dataset.

Let's first read the data using the pandas read_csv() function:

# Import libraries
import pandas as pd

# read the dataset
df=pd.read_csv('amazon_alexa.tsv', sep='\t')

# Show top 5-records
df.head()

This results in the following output:

In the preceding output dataframe, we have seen that the Alexa review dataset has five columns: rating, date, variation, verified_reviews, and feedback.

2. Feature generation using TfidfVectorizer:

Let's generate a TF-IDF matrix for the customer reviews using scikit-learn's TfidfVectorizer:

# Import TfidfVectorizer and RegexTokenizer
from...