Relationships made no sense to me when I first started using #PowerBI

I didn't have a background in IT & I didn't understand databases. I was a stats nerd. 🤓

If you're new to Power BI, here are 5 things I wish I had known about relationships when I first started:

Primary & foreign keys:

Primary key: sales table contains a SalesID for each transaction.
Foreign key: sales table contains a ProductID used for a relationship to the products table (then, within the products table, the ProductID is the primary key of that table).

One-to-many / many-to-one cardinality:

Most common relationship.

One row in Table A can match many rows in Table B.

Ex: Products sold in the sales table match a list of unique products in the products table.

Many-to-many cardinality:

⚠️ Stay away from this one in the beginning (we'll discuss why in a future thread).

Many rows in Table A match many rows in Table B.

Ex: Products sold in the sales table match a list of products containing duplicates in the products table.

Cross filter direction:

Either "single" or "both".

Filters always flow FROM the table with unique values TO the table with multiple values.

When choosing "both", ask yourself: if I filter products in the sales table, do I also want the products table to get filtered?

Inactive relationships:

Not all relationships need to be active at once.

You can specify inactive relationships for special cases where you need the inactive relationship in a DAX calculation (with the USERELATIONSHIP function) & use the active relationship in normal reporting.

5 Things I w Relationships in Power BI

❌ One of the biggest mistakes I made when I first started building #PowerBI reports:

I only used charts and colors that I thought looked cool.

👇 Do this instead

Put yourself in the shoes of the decision-maker.

Ask yourself two key questions:

1️⃣ What problem is this dashboard trying to solve?

2️⃣ What decision can be made with this dashboard?

If you don’t know the answers to these questions, ask.

Nothing will kill your project (and potential future projects) faster than a lack of purpose.

✅ Be clear on why the dashboard exists in the first place.

Avoid This Mistake in Power BI Reporting

🧹 Data Cleaning Essentials in #DataScience

Data quality matters! Let's dive into common issues and practical solutions. Starting with...

( #Python code snippets included 👇 )

Issue 1: Missing Data ❌

🔍 Problem: Missing values disrupt analysis.
💡 Solution: Imputation! Fill in gaps with mean, median, or mode. Here's a simple Python snippet:


Issue 2: Duplicates 📊

🔄 Problem: Duplicate entries skew results.
💡 Solution: Use Pandas to drop duplicates:

Issue 3: Outliers 🚀

📈 Problem: Outliers can mislead your analysis.
💡 Solution: Detect and handle outliers with the IQR method:

Issue 4: Inconsistent Data Formats 📆

🗂 Problem: Dates in various formats? Convert them to a standard format!
💡 Solution: Use Pandas to standardize date columns:

Issue 5: Text Data Cleanup 📝

🧼 Problem: Unwanted characters, HTML tags, or whitespace.
💡 Solution: Clean text data with regex and Python's string functions:

That's a wrap!

If you enjoyed this thread:
1.  Follow me @joleenbothma for more data science content
2.  Like 👍 and repost 🔄

Data Cleaning Essentials in Data Science

I spent 6 months on my own intensive data science self-study curriculum, taking my machine learning and python skills to the next level.

But it didn't go as planned.

Don't make the same mistakes I did.

🧵 👇

#DataScience #MachineLearning #Python

1. Worried too much about finding the 'best' resource

Preparation is important for self-study but I overlooked my own learning style when selecting which resources I would use.

There is no such thing as a perfect course/book and I learned this the hard way.

2. Used resources that didn't match my learning style

I am an active learner. I cannot learn by passively watching videos from an online course.

Unfortunately, most of the resources in my curriculum were online courses with only a few active learning elements.

3. Not doing enough projects

My intention was to do as many projects as possible.

But because I picked resources that were very passive, I got sucked into passive mode.

The important lesson here is to only learn as much as you need to start applying the concepts in a project

4. Fixating on learning ALL the math

Math is important to understanding machine learning.

However, you do not need to understand every single mathematical concept you come across. 

This does not make you a better data scientist. Being able to solve problems with data will.

5. Reading research papers too soon in my journey

I thought I needed to read (and understand) the research papers that were associated with the topics I was learning.

But these papers just confused and frustrated me. 

First set a good foundation, then read research papers.

That's all for now!

If you enjoyed this thread:

1. Follow me @joleenbothma for more data science content
2. RT the first tweet to share this thread with your audience

Do You Want to Self-Study Data Science? Learn From My Mistakes

My most successful business intelligence dashboards contained these key elements:

1 - Identify why we’re doing this project in the first place - how will this project help the business (be specific)?

2 - Start with the end first - identify what the end result should look like and how the project will be used once finished.

 [2/5]

3 - Identify the core metrics to display. These should be presented in a way that grabs the user's attention as fast as possible.

 [3/5]

4 - Build in layers - core metrics should be displayed front and centre. Subsequent layers can be added to the dashboard to show more and more detail as needed.

 [4/5]

5 - Driven by the user - ultimately, the project should be 100% driven by the user: the look, navigation, preferred visuals or way of representing the metrics, and level of detail.

 [5/5]

Building Successful BI Dashboards

Self-study requires a lot of discipline and motivation.

No one is there to push you - it’s all up to you.

Here are 3 of the biggest challenges I’ve faced so far in my self-study journey (+ my strategies for overcoming them).

🧵 👇

1 - Procrastination

Doom-scrolling on social media, Netflix, and video games.

These are the ways I procrastinate. They all give me a generous dose of dopamine but they don’t move me towards my goals.
 [2/8]

I have ‘micro-solutions’ to these forms of procrastination - like creating a schedule, or using the pomodoro method.

But, a longer-term solution I have found is to be clear about WHY I’m doing all this studying in the first place. This helps to keep me motivated.
 [3/8]

2 - Shiny objects

Getting an idea for a new project, going down random rabbit holes, finding a new cool app.

These 'shiny objects' can be fun and exciting and aren’t always a bad thing.

But they can make it difficult to stay focused.
 [4/8]

I have a special Notion page for shiny objects - I dump all my ideas, thoughts and interesting stuff into an ‘inbox’ that I can come back to another time.

This frees up my mind to focus on what I need to get done right now.
 [5/8]

3 - Structure

The hardest part about self-study is that there is no structure and no deadlines.

There is no one holding your hand through it all and while this can be overwhelming, it can also be the biggest benefit to self-study.
 [6/8]

I created my own curriculum with timelines to impose some structure.

But, the upside is that I get to tweak this curriculum as I go to match my own learning style and preferences.

I get to choose what I learn and how I learn it.
 [7/8]

That's it so far!

If you're interested in hearing more about my data science self-study journey then give me a follow! ☺️🙏

 [8/8]

Top 3 Data Science Self-Study Challenges

Do I even have what it takes to call myself a data scientist?

I often ask myself this question when I come across a really challenging concept, paper or project that I struggle to get to grips with.

👇 This is how I deal with these challenges and doubts

(A short 🧵)

1 - You're not alone

It’s easy to assume that everyone working in data science is a next level genius and I’m the only one struggling to understand stuff.

But that’s just not true. Everyone struggles with new concepts. The most important thing is to not give up.

2 - Go backwards.

When I can’t understand something, it’s usually because there is a hole in my knowledge that is preventing all the pieces from fitting together in my mind.

This is when I need to go back a step (or maybe a few) to build a better foundation for my knowledge

3 - Take a break.

Frustration, anger, self-doubt.

I cycle through these emotions when I really can’t figure something out.

So I just take a break for a little while (sometimes even a whole day). Then when I come back to it I am in a better frame of mind and I try again.