Subscribe|Submit a tip|Advertise with Us

📊 Welcome to BIPro 119

This week we are bringing something new to our BI and analytics readers. We have launched live AMA sessions with industry experts on the Packt DataML YouTube channel, hosted by our Growth Lead Abhishek Kaushik. These sessions give BI practitioners direct access to practical guidance, workflow clarity, and real world perspectives from experts shaping the future of Python, AI, and data driven decision making. In this edition, we highlight insights from our latest conversation with Juan Gabriel Salas, one of Udemy’s highest rated Python and data science instructors. Juan shared actionable guidance on learning Python with purpose, connecting fundamentals to real analytical workflows, strengthening data intuition, and building work that stands out in a competitive BI landscape. You’ll find the full recap below.

Knowledge Partner Spotlight: Outskill
At Packt, we’ve partnered withOutskillto help readers gain practical exposure to AI tools through free workshops, complementing the deeper, hands-on, expert-led experiences offered throughPackt Virtual Conferences.

Become an AI Genius in One Weekend
As part of their Black Friday offer, Outskill is opening access to their 2 day LIVE AI Mastermind for free. This 16 hour intensive program covers AI tools, automations, and agent building, and is rated 9.8 on Trustpilot. It is usually priced at $395, but DataPro readers can join at no cost while seats last. The live sessions run Saturday and Sunday from 10 AM EST to 7 PM EST.

Join Jason Strimpel for a hands-on workshop on December 12, where you’ll design, test, and deploy algorithmic trading strategies in Python using pandas, VectorBT, and the Interactive Brokers API. Whether you’re a quant-in-training or a finance pro, this live session takes you through the entire workflow, from identifying trading edges to executing real-world strategies.Pick a Full Pass or VIP Pass, use EARLY20 for 20% off, and reserve your seat.

Cheers,

Merlyn Shelley

Growth Lead, Packt

With an aim to improve the reader experience, we are rolling out a new AMA series in partnership with our Expert Network to bring firsthand insights directly from industry leaders in the data world.Abhishek, our Growth Lead for the data science vertical, is hosting these live AMA sessions on thePackt DataML YouTube channelto help learners gain clarity, confidence, and direction in their data journey.

This week’s session featuresJuan Gabriel Salas, a top rated Udemy instructor and founder of Frogames Formación. Juan is known for his practical teaching style, clear explanations, and deep experience helping thousands of learners master Python and build data science careers. Over the course of an hour, he shared a wide range of actionable insights on how to learn Python effectively, build strong intuition, create impactful portfolios, and navigate the fast evolving world of data science and machine learning.

This article brings together all those insights in a single narrative so that every data learner can benefit from the clarity and wisdom that came out of the session.

The Big Themes of the Session

Python fundamentals. Data workflows. Portfolio strategy. Math intuition. Future skills. And everything in between.

To make this recap easier to navigate, we have organized the insights into six major themes that emerged during the AMA.

1. Where Should Beginners Start: Python or Data Science?

The conversation kicked off with a classic dilemma. Should learners start with Python basics or jump straight into data science?

Juan’s advice was simple and rooted in experience.Start with Python, but not in a way that traps you in endless syntax drills. Instead, learn Python with a purpose. Focus on the parts of the language that matter for data work.

You do not need to become a software engineer before becoming a data practitioner. But you do need enough fluency to manipulate data, write reusable code, and understand how your tools work behind the scenes.

Python is not just another skill. It is the language that holds your entire workflow together.

What matters most at the start:

Knowing how to work with lists and dictionaries

Reading and writing files

Understanding functions and basic logic

Getting comfortable with problem solving

Once these fundamentals feel natural, transitioning into data science becomes far more enjoyable.

2. Closing the Gap Between Syntax and Real Data Science Work

Many learners understand Python basics but freeze when facing a blank project. They know what a loop is but not how to apply one in a data pipeline. They understand lists but not how to use them in a transformation.

Juan explained that this is not a lack of skill. It is a mindset shift that has not yet fully developed.

Data science is less about writing isolated lines of code and more about seeing the bigger picture. It requires thinking insystems and workflowsrather than steps.

You move from:

“How do I write this line of code?”
to

“How do I design a process that answers this question?”

You start seeing your work as a sequence of stages:
Load. Clean. Explore. Transform. Model. Evaluate. Communicate.

This shift is what turns a coder into a data practitioner.

3. How to Approach a Real Data Science Problem

One of the most valuable parts of the AMA was Juan’s breakdown of what to do when facing a new data science project from scratch. Instead of jumping into libraries or algorithms, he recommended focusing on three checkpoints that define successful work.

Checkpoint 1: Understand Data Quality

Before choosing a model or writing transformations, you must examine the data. Learn to recognise missing values, duplicated rows, strange outliers, or mislabeled categories. This is where the bulk of practical skill is formed.

Checkpoint 2: Choose the Right Modeling Approach

Is this a classification problem? A regression? A time series forecast? A clustering experiment?

Knowing what problem you are solving is more important than the model itself.

Checkpoint 3: Define Evaluation Metrics Early

Accuracy is not always the right metric. Sometimes you need precision, recall, AUC, F1, RMSE, or business specific KPIs.

Thinking about metrics early forces clarity.

Together, these checkpoints create a stable foundation for any analysis.

4. How Tiny Python Mistakes Can Break Entire ML Workflows

Juan shared a humorous yet painful story about how a single indexing mistake caused inconsistencies across an entire machine learning pipeline. The model trained successfully. The predictions looked plausible. But the results made no sense.

The culprit was a tiny transform that shifted rows.

This resonated deeply with the audience because it illustrates a truth that every data practitioner eventually learns.Machine learning magnifies small mistakes.

One misaligned array. One swapped column. One incorrect filter. One wrong assumption during data cleaning.

This is why debugging is one of the most underrated data science skills. Being patient, methodical, and curious makes you far more effective than knowing every library function.

5. What Makes a Portfolio Stand Out to Employers

Perhaps the most practical part of the AMA was Juan’s guidance on portfolio building. Many learners build beautiful notebooks that fail to impress employers simply because they focus on the wrong things.

Juan broke down the three elements that make a portfolio truly meaningful.

1. Use Real World Data

Avoid overused toy datasets. Instead, explore domains you care about. Scrape data. Gather logs. Use messy CSVs. These reveal how you work, not how well you follow tutorials.

2. Show Your Thinking

Employers want to know why you chose an approach, what decisions you made, and how you debugged problems. Narrative clarity matters just as much as code.

3. Focus on Depth, Not Quantity

Five shallow projects will not beat two strong, thoughtful, well documented projects with clear reasoning and solid results.

A strong portfolio is a story, not a showcase.

👉 Continue reading the full article on the Packt Medium handle.

See you next time!