This blog is not about the traditional roadmap to become a data analyst showing you the bare minimum of what you need. This blog is about showing you and teaching you skills with actionable examples that will make you stand out for a data analyst job or any other analyst job.
I like getting straight to the point, so let me jump into the first focus area out of the three that I will cover today, prompt engineering or in plain simple English, how to use AI effectively. To create specific and effective prompts, follow my 10-step list.
First, define a clear goal and your desired outcome. Instead of analyze the sales data, how about you use a more specific prompt like analyze the sales data and identify the top three products with the highest sales growth over the past year.
Number two in the list would be to provide relevant context and background. Don’t just say find anomalies in the data, use the prompt you are a data analyst reviewing monthly transaction data for a retail store. Identify any anomalies in transaction amounts that could indicate errors or fraud.
Number three, use positive direct instructions. For example, instead of using the prompt explain the dashboard, but don’t use jargon. Try this prompt instead. Explain the dashboard in simple terms suitable for a non-technical audience.
Number four, be specific about format, length, and style. Don’t just use the prompt summarize the customer feedback. Use the prompt summarize the customer feedback in three bullet points, each limited to one sentence. It’ll work much better, trust me.
Number five, break down complex tasks. Instead of analyze the sales data and explain any trends, use this prompt. First, calculate the average monthly sales for 2024, then identify any months where sales were 20% above or below this average and explain possible reasons.
And let’s get on with point number six. Include examples. Instead of using this prompt, write a summary for this data visualization. Use this. Example, chart, bar chart showing sales by region. Summary, the bar chart indicates that the east region had the highest sales followed by the west and south regions. Now write a summary for a line chart showing monthly website visits.
Number seven, assign a role or persona for domain-specific tasks. Explain this SQL query is not a good prompt. You are a senior data analyst explaining this SQL query to a junior team member who is new to SQL is.
Number eight, use step-by-step or chain of thought reasoning. Find outliers in the data set. Not a good prompt. Identify outliers in the data set and explain your reasoning step-by-step. It’s much better.
Number nine, iterate and test prompts. Please don’t use a prompt like, summarize this report. Do this instead, at least. Summarize this report in three sections. Key findings, supporting data, and recommendations.
And number ten, avoid ambiguity and conflicting instructions. This may sound like a no-brainer, but it’s actually not that hard to make the mistake of giving ambiguous instructions. A bad example would be, write a detailed but brief summary of the data. And a good example would be, write a concise summary of the data analysis workflow in three sentences.
Now, let me move on to the second key focus area, data visualization, because data visualization skills are one of the most in-demand skills right now, particularly through tools like Tableau and Power BI. And here are the five actions you can take right now and why you should take them.
One, master the fundamentals of data types and learn which visualization best fits each use case, such as line charts for trends, bar or column charts for comparisons, or scatter plots for relationships. This will help you simplify complex data for better decision making. Turning raw data into simple, easy to understand, actionable insights is a crucial skill.
Two, practice and practice and practice more with the popular visualization tools out there. I know the which one should I learn Tableau or Power BI debate can be overwhelming sometimes, so let me save you months of useless learning and do this. Learn only one of these tools and learn it really well. Most companies will be either Tableau or Power BI users, not both. Research the industries and companies you’d like to work for and just learn the visualization tool that the majority of them use. If that’s Tableau, learn Tableau. If that’s Power BI, learn Power BI. It’s that simple in my opinion.
Three, work on your data storytelling and communication skills. Focus on using annotations, color contrasts, and efficient layouts to guide your viewers through your visuals. Take a simple data set and create a visualization with a clear title, subtitle, and call outs explaining the so what behind your numbers.
Four, apply design best practices and seek feedback. Oftentimes, less is more. I’m a huge fan of the charts and visuals produced by The Economist. They use mostly line charts and bar charts only. Why? Because they are simple and super easy and quick to understand and interpret. They actually have an official style guide that you can check out and follow anytime. I would really urge you to have a look.
And number five, build your portfolio. Documenting your progress is one of the best ways to showcase your visualization skills. The easiest way to do this is just to publish your work online, whether you’re using Tableau or Power BI. If you want to take it a step further, write some articles that really tell the story behind your visualizations.
You can do this for free on Medium or GitHub, or you can build your own website, or you can use my Ultimate Data Portfolio if you’d like to support the channel. Last but not least, let me cover the third key focus area, cloud computing skills. These are becoming increasingly more important as more and more large corporates move their infrastructure from on-premises to the cloud.
And many startups, of course, were founded in the era of the cloud and therefore run their entire operations in the cloud. I’ve mentored hundreds of people and the biggest question people had about learning cloud computing skills is how and where to actually start. So let me answer this one today to save you the time of going down various useless rabbit holes of researching, watching tutorials, and reading random articles.
Start with choosing which cloud provider you’d like to go with, and please pick one of these three. AWS, which stands for Amazon Web Services, Microsoft Azure, or Google Cloud. They’re the biggest cloud providers in the world.
They dominate the market and you really don’t want to learn some cloud skills with a provider that very few companies use. And let me just emphasize this again, please pick either AWS, Azure, or Google Cloud because each platform has their own tools and services, hundreds of them, and naming conventions, and you need to be very familiar with the products and services on these platforms to be able to actually carry out any form of meaningful data analysis workflow. All three, AWS, Azure, and Google Cloud, have official YouTube channels with lots of useful learning that you should definitely check out.
And if you’re after something more structured and a lot quicker, then I think you would enjoy the courses I took on Datacamp. Courses like Understanding Cloud Computing, AWS Concepts, Introduction to Azure, or Introduction to GCP. And that’s all from me for today.
Feel free to continue reading more blog like this here. Thanks again for taking a little time out of your day to read this, and I shall see you in the next one.