1. Avoid Diving Straight into Analysis

Often, it's mistakenly assumed that everyone involved in the conversation is already on the same page regarding the discussion's purpose and the problem being tackled. However, this mutual understanding is frequently absent. Participants may lack context, have forgotten prior conversations, or come in with differing interpretations of the objective.

Jumping right into the analysis without clarifying the context and goals can lead to confusion later on. People's understanding of information is closely linked to how they relate it to a specific goal. If there's any disparity in how participants perceive that goal, it’s likely that subsequent information will be interpreted inconsistently, wasting precious discussion time.

I typically dedicate the first ten minutes of a one-hour meeting to clarifying context and objectives. Questions like "Why are we here?" and "What problem are we addressing?" should be articulated clearly, preferably in a single statement. Gaining consensus on this will help redirect the conversation if it strays off course later.

2. Limit Focus on Methodology and Data

As a data scientist, I often find certain methodologies or data sources thrilling and feel compelled to share them. However, this enthusiasm is not usually shared by non-data scientists, who are primarily interested in the results and implications.

It's vital to restrict discussions about data to what is absolutely necessary to bolster confidence in your methods. A brief explanation of the data source, allowing for any immediate concerns to be voiced, is usually sufficient. If there are known issues with the data, these should also be addressed concisely. Detailed discussions can be relegated to an appendix, which I find incredibly useful.

When discussing methodology, a simple mention of well-accepted methods is often enough. More in-depth details should only be provided if a custom approach was required. Focus on illustrating the core logic of your method while ensuring that you only include information necessary to build confidence in your findings.

3. Avoid Overstating Results

It’s common for data scientists to make grand claims about their findings, which can lead to misinterpretation. For instance, stating "A more diverse sales team will perform better" can be misleading without proper statistical backing. Instead, a more accurate phrasing could be "Among the sales teams we analyzed, those that performed better tended to be more diverse."

It’s crucial not to exaggerate your conclusions in headlines or summaries, as these are the elements most likely to be remembered and accepted as truth. Your work should also be defensible; if another analysis contradicts your findings, you risk damaging your credibility.

4. Simplify the Presentation of Results

Presenting extensive statistics can often confuse your audience rather than convince them. I frequently observe data scientists displaying complex tables of model statistics, only for viewers to respond with confusion, asking, "What should I be focusing on?"

If your point hinges on a specific statistic or trend, present only the relevant data in the clearest graphical format possible. The aim is for laypeople to easily see how your conclusions relate to the visuals. Avoid overwhelming them with large tables; instead, filter and highlight the most important results. Use appendices for additional, albeit less critical, information.

5. Choose the Right Communication Format

Over the years, I've discovered that the format of your communication can significantly influence how your findings are received. A key lesson is to avoid conforming to traditional slide presentations if they hinder logical flow. A strong narrative is vital for effective communication in data science.

If a fluid, linear progression is necessary for your presentation, consider using a vertical format, such as a Word document or an integrated R Markdown or Jupyter notebook. Vertical formats often serve as more effective pre-reads, allowing the audience to engage with the content without the constraints of a slide-based presentation.

These lessons have greatly improved my ability to communicate data science findings. If you have additional insights or techniques, please share them in the comments!

The first video titled "9 Common Mistakes You Shouldn't Do as a Data Scientist!" highlights frequent errors that data scientists make in their communication and analysis, providing practical advice on how to avoid these pitfalls.

The second video, "Why You Should Not Be a Data Scientist," discusses the challenges and misconceptions surrounding the profession, offering insights into the realities of the data science field.