Leading With Data· a Bicycle Guide

Partial book profile

advanced analytics for business

Saumitra N. Bhaduri

A practical guide for data scientists on applying, comparing, and deploying both classical machine learning (especially gradient boosting) and deep learning techniques to solve real-world business problems involving tabular data.

The full book profile is not available yet.

Get the book →

What it’s about

The modern world runs on tabular data—the rows and columns in spreadsheets and databases that underpin every major industry. "Advanced Analytics for Business" provides a hands-on guide to unlocking the immense value within this data using both state-of-the-art machine learning and emerging deep learning techniques. This book moves beyond theory, offering practical Python code examples that cover the entire data science lifecycle, from messy, raw datasets to deployed models in a professional cloud environment. It directly addresses the heated debate on whether gradient boosting or deep learning is better for tabular data, equipping you with the knowledge to choose the right tool for the job. You'll master best practices for feature engineering, model selection, hyperparameter tuning, and MLOps, ultimately gaining the skills to tackle real-world tabular data challenges and deliver impactful business insights.

The through-line

Who it’s for
A data scientist, analyst, or machine learning practitioner who works extensively with tabular data and wants to build powerful, effective predictive models to solve real-world business problems. They are skilled in Python and basic ML but want to master advanced techniques and understand how to choose between the competing paradigms of gradient boosting and deep learning.
The problem
Faced with complex tabular datasets, it's unclear whether to use traditional machine learning (like XGBoost) or the more hyped deep learning approaches. Choosing the wrong method leads to wasted time, suboptimal models, and difficulty explaining results. Feeling uncertain and overwhelmed by the 'controversy' and the rapid evolution of tools, doubting whether they are using the best possible approach for their problems and feeling pressure to adopt deep learning without a clear understanding of its benefits and drawbacks for tabular data.
The plan
  1. Master the fundamentals of working with tabular data, including finding, cleaning, and exploring datasets.
  2. Dive deep into classical machine learning, focusing on the state-of-the-art in gradient boosting (XGBoost, LightGBM) and best practices for feature engineering and selection.
  3. Explore the application of deep learning to tabular data, learning about various architectures (fastai, TabNet, etc.) and when they are appropriate.
  4. Learn to build and deploy end-to-end ML pipelines in a professional cloud environment (Google Cloud Vertex AI).
  5. Understand how to blend both approaches to achieve even better results.
The payoff
Confidently choose the optimal modeling approach (gradient boosting, deep learning, or a blend) for any tabular data problem. · Build high-performing, robust models that solve real business problems. · Deploy models into production using modern MLOps pipelines.

See our guide

Related profiles we’ve built

Featured in these guides

Co-cited books