6 Data Science Mistakes That Made Me Look Stupid (So You Don’t Have To)

I learned early that my love for design and pink should have little place on an analytical dashboard.

Nov 25, 2024

Data analysis can be rewarding, especially when your analysis produces expected results and is used for strategic actions.

But it can also be humbling.

Over the years, I’ve made mistakes that slowed me down and left me with my fair share of “facepalm” moments.

Mistakes that made me cringe and go, “Whaaaat?”

If you’re an experienced analyst, you can relate to these.

The good thing is these blunders became lessons that shaped my approach to analysis.

In this article, I’m sharing the top five data analysis mistakes that made me look less like a data professional and more like an amateur.

Hopefully, you can avoid these pitfalls and save yourself the embarrassment.

1. Ignoring Data Context

I have analyzed some datasets a few times without understanding their real-world context. This was common when analyzing sales data obtained from databases or online resources.

I just wanted to have some analysis to build a portfolio and show my capabilities.

What happened? A good look at the analysis interpreted customer churn as seasonal, but my dataset didn’t account for users outside a single region.

Other external factors influenced the numbers.

If this were real-life data, it would have impacted decision-making for the business.

Lesson:
Don’t start analysis without understanding where your data comes from and what it represents.

Talk to stakeholders, explore the source, and document assumptions.

This small step can save you from costly mistakes later.

Context in data analysis is a vital tool.

2. Automating Without Proper Testing

This was a painful mistake because I like automation. Simplifying things is my favorite part of doing anything.

Demystifying Automation

Once, I developed an automation pipeline to clean and transform a dataset before analysis. Eager to showcase my skills, I skipped thorough testing. Imagine my embarrassment when the pipeline failed mid-presentation and produced incomplete data outputs.

What happened?
I prioritized speed. I didn’t test the pipeline against edge cases (something that challenges the normal behavior of a system or, in this case, my automation pipeline), which led to unexpected failures when the dataset expanded.

This undermined the credibility of the analysis, but thankfully, the presentation was for my senior colleague and not an entire room of stakeholders.

Lesson Learned:
Automation is a powerful tool, but it’s only as good as its testing.

Simulate real-world scenarios, including edge cases, and always validate the output before presenting results.

Note to self: Add simple, intermediate logs to capture unexpected errors during execution.

Reliable automation is better than rushed automation.

3. Using Default Visualization Settings and Fancy Colors

Nothing says “I don’t know what I’m doing” more than using a rainbow color palette or an illegible pie chart. I want to say I have nothing against Pie charts, but this may be a lie, so let’s keep it moving.

Now, this happened to me because my other passion is design. I like colors, and I like something fanciful. Pink dashboards make me smile.

I learned early that this love for design and pink should have little place on an analytical dashboard.

After creating a dashboard that could make a beautiful design on a gorgeous evening dress, I presented it to a client with a wide grin, expecting to wow them. Only for them to squint and ask, “What am I supposed to look at here?”

I’ll leave it here for you to imagine what went through my head.

What happened?

Default settings in tools like Excel or Tableau are not always optimized for clarity and can lead to misleading or unclear charts, confuse your audience, and question your credibility. It can also extend presentation time.

Lesson:
A good visualization isn’t the best time to show your love for colors.

This can be okay for personal projects but not for stakeholders. Default settings are not your friends. Customize your visuals to highlight key insights, ensure readability, and align with the audience’s needs.

Use consistent colors and label axes, and avoid overloading your charts with unnecessary elements.

Customize visuals to suit the audience and highlight the story your data tells.

Next time, I will write on the importance of white space in dashboards and writing.

Matthew Ström’s article on “picking the least wrong colors” is one of my favorite articles on dashboard designs.

4. Partial Data Cleaning

I know what you’re thinking. Are you even ready to be a data analyst if you don’t perform data cleaning?

Early in my career, I would occasionally assume that my data was clean and perform minimal cleaning in the interest of “saving time.”

Predictably, the results? Errors and outliers that threw off the entire analysis.

Lesson:
Data cleaning is non-negotiable. Take time to remove duplicates, handle missing values, and verify the data’s integrity. It’s the foundation of every good analysis.

5. Ignoring Data Granularity

As a data scientist, one thing you will do a lot of is building predictive models.

This was a recent mistake since I am still navigating the transition to being a data scientist. So, you can call this — a rookie mistake.

I recently developed a model for a client using weekly aggregated data. My predictions seemed accurate, but the client noticed anomalies in their day-to-day operations.

It turned out that the granularity of my data was too broad, and it masked daily trends that were more critical to their decision-making.

I focused more on the high-level trends and omitted the finer details that mattered to the client’s operational goals.

Aggregating data without understanding its impact led to skewed predictions.

Lesson:
Always align your data granularity with the objectives of the analysis. Do I need detailed, hourly data, or is weekly aggregation sufficient?

Tailor your approach to match the decision-making needs of your stakeholders.

Predictive modeling is still a learning curve for me, but I am more conscious of data granularity.

6. Not clarifying stakeholder needs and input

At this point, I’m sure you’re wondering just how many mistakes I have made as a data analyst.

I have news for you. These don’t scratch the half of it.

I once delivered a perfectly polished analysis … that answered the wrong question.

This is the primary reason I always share my assumptions and metrics with stakeholders for approval and input.

In fact, a personal rule is crafting questions, metrics, and KPIs and submitting them for approval before I progress in the analysis.

This approach has saved me from solving problems that stakeholders don’t have.

Lesson Learned:
Stakeholder communication is as important as technical skills. Regular check-ins, confirming the problem statements, and making sure that the analysis aligns with their goals make you a more seasoned data professional.

Mistakes are inevitable in learning, especially in a dynamic field like data analysis. They have also taught me valuable lessons on aligning analysis with objectives.

If you make a mistake, it should not delay or stop your journey. Use it as a stepping stone to becoming a better data professional.

But you can be mindful of these common missteps, avoid the embarrassment I faced, and set yourself up for success.

The first step is that you are here, reading this article, and will definitely not repeat these mistakes.

The blunder I shared helped me improve my skills and deliver more impactful insights. The truth is, you can’t grow without making a few mistakes.

Ask yourself:

What can I learn from this?
How can I prevent it from happening again?
What steps can I take to turn this mistake into an opportunity for growth?

The most successful analysts and data scientists are not perfect — they are resilient, adaptable, and always learning.

So, the next time you have a “facepalm” moment, remember that every misstep is a chance to evolve and get better at what you do.

Now it’s your turn. Do you have a facepalm-worthy moment that shaped your journey? You can share them — because the best lessons come from shared experiences.