Posts on Ujjwal Chowdhury

ANURUP: privacy-first photo and PDF tools

Tue, 27 Oct 2026 00:00:00 +0530

Overview

Government exam, job, visa and KYC portals each want something slightly different: a 35×45 mm photo under 50 KB, a signature at 140×60 px, one merged PDF under 500 KB. ANURUP is a free set of photo and PDF tools for exactly these uploads.

It runs in the browser, so your files never leave your device, and there is no account to create. It works in Chrome, Edge, Firefox and Safari, on a computer or a phone, and you can install it as an app that keeps working offline.

All tools

Each tool has its own address, so you can bookmark the ones you use most.

Photo and signature	PDF
Passport Size Photo Maker	PDF Viewer
Resize Image to KB & Pixels	JPG to PDF Converter
Compress Image to KB	Merge PDF
Crop, Rotate & Flip Image	Split PDF
Remove Image Background	Organize PDF Pages
Signature Cleaner	Compress PDF to KB
JPG, PNG & WebP Converter	PDF to JPG / PNG
	Add Page Numbers & Watermark
	Sign & Fill PDF
	Remove PDF Password

How it works

Pick a tool on the home page, or search for the task, for example 50 KB, signature, merge or password.
Drop a file on the upload box, click to browse, or paste an image with Ctrl/⌘+V.
Change the settings and click the main button. The result shows up below with a preview.
Download it, or send it straight to another tool.

Chaining tools

A form upload usually takes more than one step. Every result has a Continue with this file row that opens the file in the next tool, so you don’t have to download it and upload it again. Some common chains:

Crop → Remove background → Passport photo → Resize to KB
Signature Cleaner → Resize to KB → Sign & Fill PDF
JPG to PDF → Compress PDF
Remove PDF Password → Merge PDF → Compress PDF

Files you pass along, or save with Save to workspace, land in the Workspace. It’s a shared tray that survives a page refresh and holds up to 60 files. Any upload box can pick from it, and you can select several files and send them to Merge PDF or JPG to PDF in one go.

Privacy

ANURUP is a static website on GitHub Pages. Every tool runs inside your browser tab with Canvas, pdf-lib, PDF.js and QPDF compiled to WebAssembly, so no server ever receives your photos, PDFs or passwords.

The page downloads three things: the app itself, web fonts from Google Fonts and, the first time you use Remove Image Background, its AI model from IMG.LY’s CDN. Your photo is still processed on your device.

The browser keeps your workspace files (in IndexedDB), your light or dark choice (in localStorage) and a copy of the app for offline use. A PDF password is used once to decrypt the file and isn’t stored anywhere. Remove PDF Password also needs the real password. It doesn’t guess or crack anything.

Install and use offline

ANURUP is a Progressive Web App. Once installed, it opens in its own window from the dock, Start menu or home screen.

Where	How
Chrome or Edge on a computer	Click Install app in the sidebar, or the install icon in the address bar
Android (Chrome)	Tap the download icon in the top bar, or Install app in the menu
iPhone and iPad (Safari)	Share → Add to Home Screen
Safari on a Mac	File → Add to Dock

After your first visit the app keeps a copy of itself (about 6 MB), and every tool works without internet, installed or not. Remove Image Background is the one exception: it downloads its AI model the first time and works offline after that. When a new version is out, ANURUP shows a Reload prompt. Nothing changes until you click it, so a tool won’t reset in the middle of a job.

Under the hood

Part	What it uses
Interface	React, TypeScript, Tailwind CSS, Radix UI
PDF	pdf-lib, PDF.js, and QPDF compiled to WebAssembly for removing passwords
Images	Canvas, and IMG.LY’s background-removal model running on the device
Build	Vite, plus a prerender step that writes a static page for every tool, a sitemap and a 404 page
Offline	vite-plugin-pwa and a Workbox service worker
Tests	Vitest
Hosting	GitHub Pages, deployed by GitHub Actions on every push to `main`

Guides

The ANURUP wiki has step-by-step guides:

Getting started: how the app is laid out and how to move a file from one tool to the next
Install as an app: add it to your dock or home screen, update it, uninstall it
Photo and signature guides: resize a photo to 20-50 KB, make a passport size photo, prepare a signature for an exam form, change a background to white
PDF guides: JPG to PDF, compress a PDF to 100 KB, merge certificates, remove the password from a bank statement or e-Aadhaar, sign a form
Workspace and chaining: keep files between tools without downloading them
Privacy and security: what stays on your device and what the app downloads
FAQ and Troubleshooting
Development: run it locally, add a tool, deploy

Found a bug? Open an issue with the tool name, your browser and what you expected to happen. Please don’t attach private documents.

NSEScraper: an open-source scraper for nseindia.com

Sat, 03 Aug 2024 17:37:11 +0000

Installation

To install using pip, use

1

pip install nsescraper

nsescraper requires Python 3 and Pandas to execute.

Features:

nsescraper package contains different functions such as:

intraday_index: Scrapes intra day data for any index from NSE
intraday_stock: Scrapes intra day data for any listed stock from NSE

Usage

Import the library:

1

from nsescraper import *

Choose a method:

To scrap current days nse index/stock data as 1 minute candle format

1

intraday_index("nifty 50")

Output :

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16


+-----+---------------------+---------+---------+---------+---------+
| | DATETIME | open | high | low | close |
+=====+=====================+=========+=========+=========+=========+
| 0 | 2024-08-02 09:00:00 | 25010.9 | 25085.6 | 24777 | 24945.8 |
+-----+---------------------+---------+---------+---------+---------+
| 1 | 2024-08-02 09:01:00 | 24946 | 24950 | 24826.8 | 24830 |
+-----+---------------------+---------+---------+---------+---------+
| 2 | 2024-08-02 09:02:00 | 24816.5 | 24816.5 | 24784 | 24792.3 |
+-----+---------------------+---------+---------+---------+---------+
| 3 | 2024-08-02 09:03:00 | 24791.5 | 24791.5 | 24777.6 | 24789.8 |
+-----+---------------------+---------+---------+---------+---------+
...
| 392 | 2024-08-02 15:32:00 | 24699.5 | 24717.7 | 24699.5 | 24717.7 |
+-----+---------------------+---------+---------+---------+---------+
| 393 | 2024-08-02 15:33:00 | 24717.7 | 24717.7 | 24717.7 | 24717.7 |
+-----+---------------------+---------+---------+---------+---------+

1

intraday_stock("britannia")

Output :

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16


+-----+---------------------+---------+---------+---------+---------+
| | DATETIME | open | high | low | close |
+=====+=====================+=========+=========+=========+=========+
| 0 | 2024-08-02 09:15:00 | 5645.9 | 5695.8 | 5645.9 | 5695.8 |
+-----+---------------------+---------+---------+---------+---------+
| 1 | 2024-08-02 09:16:00 | 5707.8 | 5732.95 | 5706.7 | 5729.85 |
+-----+---------------------+---------+---------+---------+---------+
| 2 | 2024-08-02 09:17:00 | 5715 | 5737.9 | 5709.15 | 5729.85 |
+-----+---------------------+---------+---------+---------+---------+
| 3 | 2024-08-02 09:18:00 | 5732 | 5732.85 | 5689.95 | 5697.3 |
+-----+---------------------+---------+---------+---------+---------+
...
| 400 | 2024-08-02 15:55:00 | 5720.35 | 5720.35 | 5720.35 | 5720.35 |
+-----+---------------------+---------+---------+---------+---------+
| 401 | 2024-08-02 15:56:00 | 5720.35 | 5720.35 | 5720.35 | 5720.35 |
+-----+---------------------+---------+---------+---------+---------+

To scrap current days nse index/stock data as 1 second tick format

1

intraday_index("nifty midcap 100", tick= True)

Output :

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16


+-------+---------------------+---------+-----+
| | DATETIME | Tick | 2 |
+=======+=====================+=========+=====+
| 0 | 2024-08-02 09:00:00 | 58490.4 | PO |
+-------+---------------------+---------+-----+
| 1 | 2024-08-02 09:00:01 | 58490.4 | PO |
+-------+---------------------+---------+-----+
| 2 | 2024-08-02 09:00:02 | 58490.4 | PO |
+-------+---------------------+---------+-----+
| 3 | 2024-08-02 09:00:03 | 58490.4 | PO |
+-------+---------------------+---------+-----+
...
| 15319 | 2024-08-02 15:32:59 | 57913.7 | NM |
+-------+---------------------+---------+-----+
| 15320 | 2024-08-02 15:33:00 | 57913.7 | NM |
+-------+---------------------+---------+-----+

1

intraday_stock("dr reddy", tick= True)

Output :

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16


+------+---------------------+---------+
| | DATETIME | Tick |
+======+=====================+=========+
| 0 | 2024-08-02 09:15:24 | 6891.2 |
+------+---------------------+---------+
| 1 | 2024-08-02 09:15:30 | 6886.1 |
+------+---------------------+---------+
| 2 | 2024-08-02 09:15:37 | 6900.05 |
+------+---------------------+---------+
| 3 | 2024-08-02 09:15:44 | 6900.6 |
+------+---------------------+---------+
...
| 7931 | 2024-08-02 15:58:49 | 6964.15 |
+------+---------------------+---------+
| 7932 | 2024-08-02 15:59:39 | 6964.15 |
+------+---------------------+---------+

To scrap current days nse index/stock data as required length minute candle

1

intraday_index("nifty energy", candlestick= 10)

Output :

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16


+----+---------------------+---------+---------+---------+---------+
| | DATETIME | open | high | low | close |
+====+=====================+=========+=========+=========+=========+
| 0 | 2024-08-02 09:00:00 | 44954.4 | 44954.4 | 44481.8 | 44481.8 |
+----+---------------------+---------+---------+---------+---------+
| 1 | 2024-08-02 09:10:00 | 44481.8 | 44627.7 | 44258.1 | 44625.6 |
+----+---------------------+---------+---------+---------+---------+
| 2 | 2024-08-02 09:20:00 | 44618.6 | 44684.2 | 44537.8 | 44551.7 |
+----+---------------------+---------+---------+---------+---------+
| 3 | 2024-08-02 09:30:00 | 44524.2 | 44558.8 | 44375.6 | 44404.2 |
+----+---------------------+---------+---------+---------+---------+
...
| 38 | 2024-08-02 15:20:00 | 44378.4 | 44399.4 | 44286.2 | 44297.5 |
+----+---------------------+---------+---------+---------+---------+
| 39 | 2024-08-02 15:30:00 | 44299.1 | 44334.2 | 44299.1 | 44334.2 |
+----+---------------------+---------+---------+---------+---------+

1

intraday_stock("Zomato", candlestick= 69)

To scrap historical nse index/stock data

1

historical_index('NIFTY 50')

1

historical_stock('ABB')

Brain Tumour Classification

Mon, 02 May 2022 00:00:00 +0000

Radiology is a broad subject that needs more knowledge and understanding of medical science to identify tumors accurately. The need for a tumor detection program, thus, overcomes the lack of qualified radiologists. Using magnetic resonance imaging, biomedical image processing makes it easier to detect and locate brain tumors. In this study, a segmentation and detection method for brain tumors was developed using images from the MRI sequence as an input image to identify the tumor area. This process is difficult due to the wide variety of tumor tissues in the presence of different patients, and, in most cases, the similarity within normal tissues makes the task difficult. The main goal is to classify the brain in the presence of a brain tumor or a healthy brain.

Overview

This project focuses on the classification of brain tumor images into four distinct classes: ‘glioma’, ‘meningioma’, ’notumor’, and ‘pituitary’. The primary approach involves fine-tuning the VGG16 model using the PyTorch framework. Additionally, Python’s OpenCV library is utilized for image processing, and various image augmentation techniques are applied to enhance model generalization.

GitHub Link

Technologies Used

PyTorch: The deep learning framework for model development and training.
OpenCV: Used for image processing tasks.
Python: The primary programming language for the project.

Model Architecture

The VGG16 model is employed as the base architecture for the brain tumor classification task. The model is fine-tuned to adapt to the specific requirements of the project.

Image Augmentation Techniques

To enhance the model’s robustness and improve generalization, the following image augmentation techniques are applied:

Rotation
Scaling
Horizontal and Vertical Flipping
Brightness and Contrast adjustments

Optimization Techniques

Several optimization techniques are experimented with to fine-tune the model:

Stochastic Gradient Descent (SGD): A traditional optimization algorithm.
Adam: An adaptive learning rate optimization algorithm.
RMSprop: Another adaptive learning rate optimization algorithm.
AdaGrad: Adaptive gradient optimization algorithm.

Experimentation and Tuning

The optimization parameters for each technique are carefully tuned to achieve optimal performance. The model undergoes thorough experimentation to identify the best combination of hyperparameters and optimization algorithms for the given task.

Conclusion

The Brain Tumour Classification project demonstrates the successful implementation of a fine-tuned VGG16 model for accurately classifying brain tumor images into four classes. The combination of image augmentation techniques and experimentation with various optimization algorithms contributes to the overall success of the project.

For detailed implementation code and results, refer to the project repository and notebooks.

Photo by Wikipedia

Forecasting and Anomaly Detection

Mon, 27 Jun 2022 00:00:00 +0000

GitHub Repo

In the “Forecasting-and-Anomaly-Detection” project, we delved into the realm of machine learning to enhance our understanding of Walmart sales data. The project focused on both time series forecasting and anomaly detection, employing a diverse set of techniques to achieve accurate predictions and identify unusual patterns in the sales data.

Forecasting Techniques:

Regression Techniques

Polynomial Regression: Leveraging the flexibility of polynomial functions to capture non-linear relationships in the data.
Multiple Linear Regression: Utilizing multiple features to model the sales data and uncover complex dependencies.
Lasso Linear Regression: Introducing L1 regularization to encourage sparsity in the model, selecting only the most influential features.
Ridge Linear Regression: Applying L2 regularization to prevent overfitting and improve the generalization of the model.
Elastic Net Regression: Combining L1 and L2 regularization to harness the strengths of both techniques.

Time Series Forecasting

ARMA (AutoRegressive Moving Average): Employing ARMA models to understand and predict the temporal patterns in Walmart sales data.

Anomaly Detection Techniques:

Anomaly Detection Techniques using Unsupervised Learning Approaches

KNN Regressor: Utilizing the K-nearest neighbors algorithm to detect anomalies based on the deviation from the expected sales values.
DBSCAN (Density-Based Spatial Clustering of Applications with Noise): Identifying anomalies by clustering the sales data and isolating points that do not conform to any cluster.
LSTM with Autoencoders: Leveraging Long Short-Term Memory networks combined with autoencoders for capturing complex temporal dependencies and detecting anomalies in the sales data.

Data Version Control (DVC)

To maintain the integrity and traceability of our project, we adopted the Data Version Control (DVC) Python library. DVC facilitated the versioning of our data, ensuring reproducibility and providing a robust framework for collaboration.

This project represents a holistic exploration of machine learning techniques for forecasting and anomaly detection, combining traditional regression methods with cutting-edge time series forecasting and unsupervised learning approaches.

Paris Summer Olympic Games 2024 Dashboard

Wed, 27 Nov 2024 06:18:40 +0000

Olympics 2024 Data Dashboard

GitHub Repo

Project Overview

This Power BI dashboard provides an interactive visualization of the Olympics 2024 data, focusing on key statistics related to medals, athletes, countries, and historical performances. The dashboard is divided into five pages, each tailored to present specific insights through dynamic filtering and visually engaging elements.

Dataset Information

Source: Kaggle - Paris 2024 Olympic Summer Games Dataset
Description:
The Paris 2024 Olympic Summer Games dataset offers comprehensive data about the Summer Olympics held in 2024. It includes:
- Participating countries
- Athletes
- Sports disciplines
- Medal standings
- Key event details
  The dataset provides a solid foundation for analyzing the event’s performance and trends.
Additional Resources:
- Official Site - Paris 2024 Olympics
- Wikipedia - 2024 Summer Olympics

Pages Overview

1. Home

Purpose: The landing page of the dashboard.

Features:
- Navigation buttons or links to seamlessly guide users to other pages.

2. Overview

Purpose: Provides a comprehensive summary of medal statistics for the Olympics 2024.

Features:
- Medals by Country and Gender: Visualized using bar/column charts or other relevant visualizations to showcase the distribution of medals.
- Key Highlights:
  - Created using a Measure Query to dynamically calculate and display critical statistics (e.g., total medals won, top-performing countries).
- Custom Map Visualization:
  - Displays medallist density by country, offering a geographic perspective.
  - Enhances understanding of country-wise performance.
- Game-wise Filtering:
  - Users can filter all visualizations on this page by selecting specific games.
  - Improves interactivity and customization for detailed analysis.

3. Athletes

Purpose: Focuses on individual athletes and their achievements.

Features:
- Filters:
  - By Country, Game, Gender, and Medals Won.
  - Enables users to drill down into specific athlete data based on selected criteria.
- Visualizations:
  - Highlight athletes’ contributions to their countries’ performances.
  - Possible use of tables, cards, or slicers for detailed and filtered views.

4. Country

Purpose: Analyzes the performance of countries in the Olympics 2024.

Features:
- Summary Statistics:
  - Total medals won by countries.
  - Comparative analysis of countries’ performances.
- Game-wise Filtering:
  - Allows users to explore country-level statistics for selected games.
- Visualizations:
  - Incorporates bar charts, tables, or other visual elements for an impactful representation.

5. Historical

Purpose: Explores historical Olympic games data and trends.

Features:
- Summary Statistics:
  - Historical medals won by countries.
- Filters:
  - By Country and Games, enabling users to focus on specific historical data.
- Visualizations:
  - Trends over multiple Olympic years to understand performance changes.
  - Graphs and timelines to depict historical achievements effectively.

Key Highlights

Interactivity: Filters across pages allow for granular analysis and customization.
Custom Mapping: A unique visual element that enhances geographic insights.
Dynamic Measures: Queries calculate and display real-time statistics.
Comprehensive Visuals: Diverse charts, maps, and tables present data intuitively.

Applications

Suitable for analysts, sports enthusiasts, and stakeholders interested in Olympics 2024 performance data.
Helps in understanding medal trends, athlete achievements, and country-wise statistics.
Provides a historical perspective to assess long-term performance.

Visualizing Microsoft AdventureWorks Data

Mon, 25 Nov 2024 09:47:17 +0000

GitHub Repo

AdventureWorks Sales Analytics Dashboard

Project Overview

This Power BI dashboard provides comprehensive sales analytics for AdventureWorks, offering insights into revenue, orders, returns, and customer demographics. The dashboard consists of three interconnected pages, each focusing on specific aspects of the business.

Dashboard Structure

1. Sales Overview Page

Purpose: Provides a high-level summary of key business metrics with dynamic filtering capabilities.

Key Features:

Revenue analysis across different dimensions
Order volume tracking
Returns monitoring
Product performance metrics

Filters:

Order Category
Order Subcategory
Geographical Location
Date Range

2. Performance Metrics Page

Purpose: Compares current performance against targets and tracks temporal trends.

Key Visualizations:

Current month vs target comparison for:
- Order volume
- Revenue
- Returns
Weekly profit and revenue trends
Performance gauges and KPI cards

3. Customer Analytics Page

Purpose: Deep dive into customer demographics and behavior patterns.

Customer Segmentation by:

Gender distribution
Income levels
Occupational categories
Age groups
Order history

Data Source

Microsoft AdventureWorks sample database

Technical Implementation

Platform: Microsoft Power BI

Interactive Features

Cross-filtering capabilities across all pages
Drill-down functionality from category to subcategory levels
Dynamic date range selection
Tooltip enriched visualizations

Intended Users

Sales Management
Executive Leadership
Regional Managers
Product Teams
Customer Service Representatives

Business Benefits

Informed Decision Making
- Real-time access to critical business metrics
- Comparison of actual performance against targets
Customer Insights
- Better understanding of customer segments
- Identification of high-value customer groups
- Analysis of purchasing patterns
Operational Efficiency
- Quick identification of underperforming areas
- Monitoring of returns and potential quality issues
- Geographical performance analysis

Future Enhancements

Integration with predictive analytics
Addition of inventory management metrics
Enhanced customer segmentation analysis
Mobile-optimized view

Usage Guidelines

Navigating the Dashboard
- Use the tabs at the bottom to switch between pages
- Apply filters consistently across all pages
- Utilize drill-through functionality for detailed analysis
Best Practices
- Start with the Overview page for high-level insights
- Use date filters to focus on specific periods
- Export specific views for reporting purposes

Starbucks Drinks Dashboard

Tue, 26 Nov 2024 04:27:13 +0000

Starbucks Drinks Nutritional Dashboard

GitHub Repo

Project Overview

A comprehensive Power BI dashboard analyzing Starbucks drink products, focusing on nutritional content, health insights, and store location information. The dashboard provides a holistic view of Starbucks’ product offerings through four interconnected pages.

Key Visualizations:

Total number of drink products
Drink categories distribution
Average nutritional values across product lines
Top-performing drinks by various nutritional criteria

Dashboard Architecture

1. Overview Page

Purpose: Provide a introduction page of Starbucks drink portfolio and key nutritional metrics.

2. Nutritional Insights Page

Purpose: Detailed nutritional breakdown of Starbucks products

Nutritional Analysis:

Vitamin content visualization
Protein levels across different drinks
Fat content analysis
Vitamin C concentration
Interactive comparisons between products

Key Features:

Comparative charts
Drill-down capabilities
Filtering options by drink type and size

3. Advanced Nutritional Insights Page

Purpose: In-depth exploration of nutritional challenges and health-critical metrics

Specialized Visualizations:

Trans Fat Content Analysis
- Identification of high trans fat products
- Comparative trans fat levels across drink categories
Sugar Content Mapping
Mineral composition
Calorie density visualization

Health Focus:

Warnings for high-risk nutritional profiles
Color-coded indicators for nutritional extremes

4. Store Locator Page

Purpose: Store location mapping

World map with Starbucks store locations

Technical Implementation

Platform: Microsoft Power BI

Data Processing:

Advanced data cleaning
Nutritional value normalization

Interactive Dashboard Capabilities

Cross-page filtering
Dynamic data exploration
Responsive design

Target Audience

Health-conscious consumers
Nutritionists and dieticians
Starbucks management
Product development teams
Marketing strategists

Business Insights

Nutritional Transparency
- Comprehensive product information
- Clear health impact visualization
Product Strategy
- Identify high-performing nutritional profiles
- Guide product development
- Support marketing positioning
Consumer Education
- Empower informed beverage choices

Future Enhancement Roadmap

Integration with real-time nutritional tracking
Personalized nutritional recommendation engine
Expanded global store data

Usage Guidelines

Navigation
- Use page tabs for different perspectives
- Utilize filters for focused analysis
- Hover for detailed tooltips
Interpretation Tips
- Compare drinks within categories
- Use store locator for nearby options

Data Disclaimer

Important Notice:

🚨 Data Integrity Warning 🚨

This dashboard is created using DUMMY/SIMULATED DATA.
The nutritional information, store locations, and all visualizations are NOT REAL.
Data is purely for demonstration and educational purposes.
DO NOT use this data for actual nutritional planning or decision-making.
All data points are synthetic and do not represent actual Starbucks products or nutritional facts.

Analysis on Complete Works of Swami Vivekananda

Fri, 08 Dec 2023 00:00:00 +0000

Github Repo

This project aims to apply the Flajolet-Martin Algorithm to estimate the cardinality of unique elements in the entire collection of Swami Vivekananda’s complete works. The Flajolet-Martin Algorithm is a probabilistic method commonly used for large-scale data sets, providing an efficient way to estimate the number of distinct elements without needing to store them explicitly.

Objectives

Implement the Flajolet-Martin Algorithm in a scalable manner suitable for processing the extensive corpus of Swami Vivekananda’s works.
Develop a robust data processing pipeline to handle the text data and generate the necessary input for the algorithm.
Fine-tune the algorithm parameters and validate its accuracy against known cardinality benchmarks to ensure reliable estimates.

Technologies and Tools

Programming Language: Python
Data Processing: Pandas, Spark
Flajolet-Martin Algorithm Implementation: Custom Python code
Version Control: Git
Documentation: Markdown

Expected Outcomes

A scalable and efficient implementation of the Flajolet-Martin Algorithm tailored for the unique characteristics of Swami Vivekananda’s complete works.
Accurate cardinality estimates for the distinct elements in the dataset.
Documentation detailing the project methodology, implementation details, and findings.

Future Work

Potential future enhancements could include exploring other probabilistic algorithms for cardinality estimation, optimizing the algorithm further, or extending the analysis to specific subsets of Swami Vivekananda’s works.