https://github.com/rachakondaganesh/using-nlp-online-and-retail-order-review-project
analyzed customer reviews from online and retail orders. Performed sentiment analysis, keyword extraction, and topic modeling to identify trends, satisfaction drivers, and pain points. Used Python (NLTK, spaCy) and visualization tools to present actionable insights for improving customer experience and product strategy.
https://github.com/rachakondaganesh/using-nlp-online-and-retail-order-review-project
bivariate-analysis lambda matplotlib-pyplot nlp-machine-learning pandas sent-tokenize stemming stopwords trivariate-analysis unicode-data univariate-analysis word-tokenization
Last synced: 27 days ago
JSON representation
analyzed customer reviews from online and retail orders. Performed sentiment analysis, keyword extraction, and topic modeling to identify trends, satisfaction drivers, and pain points. Used Python (NLTK, spaCy) and visualization tools to present actionable insights for improving customer experience and product strategy.
- Host: GitHub
- URL: https://github.com/rachakondaganesh/using-nlp-online-and-retail-order-review-project
- Owner: Rachakondaganesh
- Created: 2025-10-08T11:18:22.000Z (28 days ago)
- Default Branch: main
- Last Pushed: 2025-10-08T11:24:52.000Z (28 days ago)
- Last Synced: 2025-10-08T13:26:34.233Z (28 days ago)
- Topics: bivariate-analysis, lambda, matplotlib-pyplot, nlp-machine-learning, pandas, sent-tokenize, stemming, stopwords, trivariate-analysis, unicode-data, univariate-analysis, word-tokenization
- Language: Python
- Homepage:
- Size: 6.4 MB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# 🛍️ Online Retail Order Reviews NLP Project
This project performs **Natural Language Processing (NLP)** on a large dataset of **online retail product reviews** written in **Portuguese**.
The goal is to analyze customer sentiments through **review scores** and text mining techniques like **unigrams**, **bigrams**, and **trigrams**.
It also visualizes rating distributions and frequent word patterns to better understand **customer satisfaction** and **feedback trends**.
---
## 📘 Overview
This dataset contains over **100,000 product reviews** from an online retail platform.
The analysis focuses on:
* Understanding **review score distribution** (1–5 stars)
* Cleaning and normalizing **Portuguese text**
* Tokenizing text and removing **stopwords**
* Generating and analyzing **N-grams** (Unigrams, Bigrams, Trigrams)
* Visualizing frequent patterns for positive (5-star) and negative (1-star) reviews
---
## 🧠 Key Insights
* 📊 **5-star reviews** dominate the dataset, indicating overall customer satisfaction.
* ⚠️ **1-star reviews** show recurring issues related to delivery delays and product quality.
* 🔠 Top unigrams and bigrams reveal common terms customers use in both positive and negative feedback.
* 🗣️ Portuguese-specific stopwords were removed to retain only meaningful words.
* 🎨 Visualizations provide comparative frequency distributions for **positive** and **negative** review language.
---
## ⚙️ Tools & Technologies
| Tool | Purpose |
| ----------------------------------- | ----------------------------------------------------- |
| **Python** | Core programming language |
| **NLTK (Natural Language Toolkit)** | Tokenization, stopword removal, and N-gram generation |
| **Pandas** | Data manipulation and handling |
| **Matplotlib & Seaborn** | Data visualization |
| **UnicodeData** | Removing text accents for normalization |
| **Portuguese Stopwords** | Filtering non-informative words from reviews |
---
## 📊 Project Workflow
### 🔹 1. Data Loading & Preprocessing
* Load dataset: `order_reviews.csv` (≈100,000 rows)
* Convert `review_creation_date` to datetime format
* Visualize **review score distribution** using pie charts and bar plots
### 🔹 2. Text Cleaning
* Convert all text to lowercase
* Remove accents using `unicodedata.normalize()`
* Tokenize text using `nltk.word_tokenize()`
* Remove Portuguese stopwords using NLTK’s corpus
### 🔹 3. N-Gram Generation
Generate:
* **Unigrams** → single words
* **Bigrams** → two-word combinations
* **Trigrams** → three-word combinations
Used for both:
* **5-star reviews** (positive sentiment)
* **1-star reviews** (negative sentiment)
### 🔹 4. Frequency Distribution Visualization
Plotted top 25 most frequent **n-grams** for both positive and negative reviews using:
```python
nltk.FreqDist(tokens).plot(25, cumulative=False, color=color)
```
---
## 📂 Project Structure
```
Online_Retail_Order_Reviews/
│
├── order_reviews_analysis.py # Main Python analysis script
├── order_reviews.csv # Dataset (100k+ Portuguese reviews)
├── README.md # Project documentation
└── images/ # (Optional) visualizations and n-gram plots
```
---
## 🚀 How to Run
### 🔹 Prerequisites
Install all dependencies:
```bash
pip install pandas matplotlib seaborn nltk
```
Also download NLTK datasets (only once):
```python
import nltk
nltk.download('punkt')
nltk.download('stopwords')
```
### 🔹 Execute the Script
```bash
python order_reviews_analysis.py
```
* Pie charts and bar plots display review distributions
* N-gram frequency charts visualize word usage patterns
---
## 📈 Visualizations
1. **Pie Chart:** Distribution of review scores (1–5)
2. **Bar Graph:** Count of reviews per score on a dark grid background
3. **N-gram Frequency Graphs:**
* Top 25 Unigrams for 5-star reviews
* Top 25 Bigrams for 1-star reviews
* Top 25 Trigrams for both categories
---
## 💡 Future Enhancements
* Add **sentiment polarity analysis** using TextBlob or Vader
* Translate Portuguese text for global reporting
* Implement **topic modeling (LDA)** for deeper insight into customer issues
* Deploy a **Streamlit NLP dashboard** for interactive visualization
* Integrate **word clouds** for better visual storytelling
---
## 👤 Author
**Rachakonda Ganesh**
📧 [[rachakondaganesh60@gmail.com](mailto:rachakondaganesh60@gmail.com)]
🔗 [GitHub](https://github.com/Rachakondaganesh)
🔗 [LinkedIn](https://www.linkedin.com/in/rachakonda-ganesh-2782452a8)
---
## 🏁 Conclusion
This project highlights the power of **NLP in retail analytics**, converting unstructured text data into meaningful insights.
By combining **Python**, **NLTK**, and **visualization tools**, it offers a data-driven view of customer satisfaction and recurring pain points — essential for enhancing user experience and improving product quality.