Stack Overflow Text Classification

Multi-Class Classification on Stack Overflow Questions
This project involves building a multi-class classification model to predict the tag of a Stack Overflow question based on its content. The goal is to classify programming questions into specific tags such as Python, CSharp, JavaScript, and Java using Natural Language Processing (NLP) techniques.
Key Features:
- Text Classification: Uses NLP to analyze programming questions and automatically categorize them into one of four tags.
- Multi-Class Classification: Handles tasks where each question belongs to one of several possible classes.
- Data Preprocessing: Cleaned and tokenized thousands of questions, normalizing and transforming text for model training.
- Model Architecture: Built with TensorFlow, optimized for text input, trained on labeled question data.
- Evaluation and Prediction: Measured performance using accuracy and predicts the correct tag for unseen questions.
Technologies Used:
- Python – Core programming language for data handling and model building
- TensorFlow/Keras – For deep learning model development
- Pandas & NumPy – For data manipulation
- Matplotlib – For visualizations and performance tracking
Business Impact:
- Enhanced User Experience: Helps users find answers faster by automatically tagging questions.
- Improved Search Functionality: Better categorization improves filtering and search results.
- Content Management: Reduces manual tagging workload, enabling scalable growth of user-generated content.
Code Snippet
import tensorflow as tf
import os
import re
import shutil
import pandas as pd
import numpy as np
import string
import pandas as pd
import matplotlib.pyplot as plt
# Load the dataset
url='https://storage.googleapis.com/download.tensorflow.org/data/stack_overflow_16k.tar.gz'
data=tf.keras.utils.get_file('stack_overflow_16k',url,untar=True, cache_dir='.',
cache_subdir='')
# Preprocessing
def standadization(input_data):
lower_data=tf.strings.lower(input_data)
replace_data=tf.strings.regex_replace(lower_data,'<br />','')
return tf.strings.regex_replace(replace_data,'[%s]'% re.escape(string.punctuation),'')
# Model Architecture
model=tf.keras.Sequential()
model.add(tf.keras.layers.Embedding(1000,16))
model.add(tf.keras.layers.Dropout(0.2))
model.add(tf.keras.layers.GlobalAveragePooling1D())
model.add(tf.keras.layers.Dropout(0.2))
model.add(tf.keras.layers.Dense(4))
model.summary()
# Make predictions
export_model =tf.keras.Sequential([vectorize_layer,model,tf.keras.layers.Activation('softmax')])
export_model.compile(
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False), optimizer="adam", metrics=['accuracy']
)



















