Eric Weber

Years of experience *seems* like a reasonable metric. It’s not, at least in a vacuum. But what’s the alternative? Ask 5 things about what those *years* entail:

1. What did you learn during those years that make you more impactful to an organization?

2. What value did you add to the company in those years? Did you provide game changing results?

3. Were you promoted in those years? Why? Speak to how you accelerated within your role.

4. What did you learn in those years that someone with fewer years could not easily replicate?

5. Think about what you would be without those years of experience. What would you know?

Problems are easy to find. Solutions are harder.

What would you do?

#data #datascience

Company: “We love data. We have lots.”
Analyst: “Here’s what the data reveals.”
Company: “We don’t agree with that.”
Analyst: “I thought you loved data.”
Company: “Not that data.”

It’s easy to say data is valuable when it agrees with company priority and direction.

It’s so much harder when it goes against long held beliefs.

Don’t despair! Look at the situation as an opportunity to learn and to help others learn. Expect pushback. But work with them as a partner not a blocker. You’ll be surprised at the outcome!

#datascience

Data cleaning is *brutal*. I found a column where NULL values were coded as NA, 999, 12345, “blank”, ___, [], and “oops”. What I’ve learned about data cleaning:

1. 80 percent of a data professional’s time on data cleaning may be an underestimate.

2. Anyone who says the data is probably good to go definitely hasn’t checked it. Or did and is passing the “fun” onto you.

3. “Just pull the data for our meeting” means “please do 12 hours of data cleaning in the next 30 minutes but pretend it wasn’t bad”.

4. You really need to know how the data was collected and have industry knowledge to clean it effectively. Most times we don’t.

5. If you scream inside your head at the data messiness people don’t look at you as weirdly as when you scream at it out loud.

Data is fun, isn’t it? What’s your favorite data cleaning story?

#data #machinelearning #datascience #analytics

Learning SQL can be tough because there’s a lot to learn. I trust these 10 resources most. I’ve used each and found them to be accurate and user friendly:

1. Zachary Thomas' SQL Questions https://lnkd.in/g-JJzuD
2. Select * SQL: https://selectstarsql.com/
3. Leetcode: https://lnkd.in/g3c5JGC
4. LinkedIn Learning: https://lnkd.in/gQXFc4n
5. Window Functions: https://lnkd.in/g3RtPCJ
6. HackerRank: https://lnkd.in/grv_9sB
7. W3 Schools: https://lnkd.in/gJPfrrv
8. CodeAcademy: https://lnkd.in/gT5xmpN
9. SQLZoo: https://sqlzoo.net/
10. SQL Bolt: https://sqlbolt.com/

Note 1: I’ve used each of these, some more than others. Leetcode tends to be my go to but each has tons of value.

Note 2: There are likely other resources. That’s great. I just haven’t gotten to all yet!

Note 3: Not all of these are free. If I’m including a paid version it’s because I think it’s worth it.

#datascience #data #sql #analytics

I used to be intimidated by SQL interviews. Good news! There are only a few concepts that show up most often. Here are the 15 concepts I’ve seen a lot:

1. CASE WHEN. Shows up all the time.
2. Self joins. Common in product.
3. DISTINCT and GROUP BY
4. Left vs outer joins.
5. UNION. Rarely discussed but frequent.
6. SUM and COUNT
7. Date-time manipulation
8. String formatting, substring
9. Window functions like rank and row
10. Subqueries
11. HAVING vs WHERE
12. In some cases LAG and LEAD
13. Understanding indexing
14. Running totals
15. MAX and MIN

While the SQL world is huge and overwhelming if you focus on these and related topics you’re well on your way to getting ready!

What did I miss?

#data #datascience #sql

Wow! Sometimes data visualization fails are their own art. I have *no* idea what mistake led to this chart. Tried to recreate it in Excel. No luck. Ideas?

1. The colors are wildly strange. Appear to be random 🤷‍♂️

2. The bar sizing. How does this happen? It is amazingly confusing.

3. I thought maybe it was upside down but that doesn’t account for all of the issues.

4. How does this make it on TV? Did no one look at it first?

5. I wish they had more than 5 days of data to make it easier to figure out what’s going on here.

Let me know if you solve it!

#data #datascience #datavisualization

Python, SQL and Tableau only get you so far. The story you tell matters most. Its through the story that you deliver real, visible *value*. 5 things I've learned:

1. The tool is only as good as the person and their understanding of the context around it. Context and domain matters - from data cleaning, to modeling to interpretation.

2. Data insights must be actionable and interpretable. I don’t care how you made it, if you build something that doesn’t allow for decision or insight it missed the mark.

3. We have to sell. Our ideas, our models and our conclusions. We have to make it obvious that what we did is connected to business. Don't assume this just is understood.

4. The best tools in the world don’t fix a bad problem definition (or lack of one). Knowing what problem to solve is a very context driven and business dependent decision.

5. People are still the core of what we do. They have emotions and biases. No tool, data or technique can overcome that. We have to respect that and work around and with people!

What's your best tip for keeping focused on delivering value?

#data #machinelearning #datascience #analytics