Kevin Adams

I’m often asked by friends and family what it means to be a Data Analyst at Kroger. In order to convey a sense of what it means to work with ‘Big Data’, I often use the following analogy.. Imagine you go down to your local grocery store and buy $200-$300 worth of grocery. A pretty good haul for your household. Your receipt is usually pretty long, about as long as your arm. Now imagine working with all the invoices for all the items that come into the store, and tracking sales of those items for every customer at 3,000 stores across America. Welcome to my world. UPC codes (Base, Case and Consumer) .. delivery invoices.. costs.. credits.. etc. Things delivered directly and items from the warehouse.. But with an ocean of data, just as in the real ocean, the properties of the data (just like those of water) don’t necessarily change. You just need better tools and lots of patience. SQL, R, Python and good old trusty Excel combined with the Microsoft Azure Cloud and Power Platform are the tools that help me navigate this data ocean and chart a course for success.

18

8 Comments

Want to ditch data analysis confusion & become a rockstar? Our latest post defines reproducibility & shows you how to document your work for clear, repeatable results. https://wix.to/kEpcU5r #newblogpost #DataScience #Reproducibility #dataanalystskills #analytics

2

1 Comment

This was my first time using KML files within Tableau as using map data isn't in my day-to-day. It was surprisingly straightforward to implement and does a good job helping to visualize flows between two points on a map. https://lnkd.in/gDVhz6uq #MakeoverMonday #dataviz #datafam

15

Usually, when I discuss self-service analytics, I come up with more reasons why I'm not a fan. For my latest rant, I noticed a trend when self-service is discussed. I hear a standard set of use cases for self-service. "Would it be nice if analysts could look up for themselves...." ➢ Revenue by quarter ➢ Top customers ➢ Top selling regions/locations. If your company still needs answers to these, you have a problem. These basic use cases were solved years ago, and any basic selling platform can provide them out of the box. Should we have a data team working on building out a platform that answers questions used in vendor demos? If we spend so much time and resources on building data platforms/ assets/ products, we should work on the most valuable use cases. Allowing anyone in the business to create anything because of their curiosity is not a valid use case. Also, consider that the examples (which are usually revenue/sales-based) I mentioned above are outcomes. They should be used to measure the effectiveness of a business initiative. ➢ New product launch ➢ Entering a new market ➢ Reach out to a new customer base ➢ New marketing campaign ➢ Revised customer retention strategy This is when we leave the self-service space. What are the corporate initiatives that business functions are working on, and performance is measured against them? How does the data platform/ assets/ products you're building help with them? As a data team, wouldn't you rather demonstrate how your work aligns with the business initiatives of your business partners?

14

8 Comments

Context is everything when building integration. If you don’t know *why* you are alerting people (*who*) to something that happened (a new account, a data anomaly, a missing step in a process) and *what* they need to do next, you’ll have a hard time determining *when* they need to do it and *how* to complete it. These facts sound a lot like the same steps to take to write a sound journalistic pyramid. It’s also why one-shot AI/ML efforts without guardrails are likely to be wrong. Focus on the context, think about what a human would do, and then you’ll be on track to taking a piece of the process and automating it.

1

2 Comments

When managing data teams, I didn't love weekly 30 minute 1:1's with my reports. I preferred to do a 1:1 every two weeks. Generally 1 hour, sometimes 30 minutes, depending on the person / their role. The weekly meetings turned into tactical status updates that could have been emails. I found doing a 1:1 every two weeks let us cover meatier topics in more depth. I trusted my team to handle their work independently week-to-week. And you should be talking outside meetings :) It slows things down when, instead of sending a quick note, people wait two or three days until their next 1:1. Thought I'd share since there have been a lot of articles floating around about how the Nvidia CEO doesn't do 1:1's. If you don't love standard 1:1's, there are things you can do besides just getting rid of the meeting. I think it's important that people get some dedicated time with their managers. It really depends on your team (the 1:1 is more for their benefit than yours) but try mixing things up!

88

17 Comments

Another Monday means its another opportunity to practice using Tableau. There isn't anything super crazy going on with my viz this week but I tried to approach the problem in a more interactive way. You can input your income and see what you should be spending on an engagement based on what your peers are spending. Any feedback is appreciated. https://lnkd.in/gZysusgy #MakeoverMonday #dataviz #datafam

5

Just wrapped up delivering “Automations for Educatiors” with Leon Furze In it, we covered why automations are something to get excited about - and how they level up from simply using tools like ChatGPT. We walked through some great examples including 👉🏼 a custom GPT that can handle loads of information, process it, give meaningful ideas, and then act on your ideas 👉🏼 bots that can walk both staff and students through essential materials (think course materials, onboarding, HR, training and more) 👉🏼 an automation that can take an entire new syllabus and turn it INSTANTLY into lessson plans, resources and equipment guides and time-stamped plans, PowerPoints and images, oh and even some podcasts just for fun This is where AI gets exciting 🎉 As one participant said “If you can describe the process, you can delegate it to AI…that’s a big take home for me” 👏👏👏 It’s about delegating work so teachers and staff can do more of what matters most. If you didn’t manage to grab a ticket live, comment “webinar” in the comments and we will send you the recording. #ai #aiineducation #aiautomations

32

67 Comments

Had fun playing around with some of the newer Tableau features. Specifically, I was working with the dashed lines introduced a while back and learned a couple of tricks with regards to labeling the endpoints of this viz. https://lnkd.in/gXxMru38 #datafam #MakeoverMonday #dataviz

2

Do you speak data? The ability to be data literate is critical for everyone to progress in their career. Gartner reports that 82% of business leaders expect all of their employees to be data literate. That is why I was excited to speak on this topic to Best Buy's Emerging Careers ERG. Data Literacy is usually defined as “the ability to read, write and communicate data in context." It was exciting to experience the energy and enthusiasm of almost 200 young professionals eager to learn. Too often people misuse data by taking it out of context (if you are looking for an example listen to most politicians) - and I am hopeful that these future leaders at Best Buy will help transform our culture to making data-driven decisions. Data is often hyped as the "new fuel" or the "new currency" - but it is worthless if everyone does not know how to speak the same language. Thanks to Soren and the group for inviting me in. How are you helping your company speak the language of data? #dataliteracy #bestbuy #cdo #datafluency #datagovernance #cdo

53

6 Comments

Very enlightening discussion on AI. Badly labeled data: “And so the training data has 50% error. But if you train up backpropagation, it gets down to 5% error or less. In other words, from badly labeled data, it can get much better results. It can see that the training data is wrong. So these big neural nets can actually do, they can do much better than their training data, and most people don't realize that.” GPU: “and initially we just got gaming GPUs and discovered they make things go 30 times faster, and then we bought one of these Tesla systems with four GPUs and we did speech on that and it worked very well, and then in 2009 | gave a talk at NIPS and I told a thousand machine learning researchers, you should all go and buy NVIDIA GPUs” Pick research topic: “But my theory is that I look for something where everybody's agreed about something and it feels wrong.” Next compute: “So instead of using like a megawatt, we could use like 30 watts like the brain, and we could run these big language models in analog hardware.” Time scales: “So one big thing that we still have to catch up with neuroscience on is the time scales for changes. You just have two timescales in neural nets. In the brain, there's many timescales at which weights change.”

1

Machine Learning Week conference is in Phoenix, AZ this week where the forecast is 110+. Now we all know correlation does not mean causation, but when it comes to AI & ML this conference really brings the heat! Come join the fun! #machinelearningweek #aiplaybook

75

8 Comments

A fun #datavisualization that doesn’t get much attention. I think it visualizes distributions in an easier to read format than box-and-whisker plots alone.

4

In your Analytics career you can choose between 1. working within a single industry or domain, or 2. jumping around between different industries There are advantages to both! Sticking to one industry: - You probably get more replies when applying - You hit the ground running faster in a new job - Your domain knowledge compounds over time Jumping around: - Wider pool of job openings available to you - Get exposed to a wider variety of tools & methods - On average, it helps you grow more as a generalist Based on your experience, what else would you add? And what do you prefer? #data #analytics #career

101

28 Comments

Stakeholder: "We need this for the board deck this afternoon." Analyst: "I'm on it" Analyst thinks: Why did you wait until last minute to tell me about this? Analyst drops everything to get the data needed for the request. Next day.... Analyst: How did the board meeting go? Stakeholder: We ran out of time and never got to it. meanwhile... 3 other expectations were missed due to the urgent need. Sound familiar? Try these tactics: 1. Get the board meeting CALENDER from the CEO's EA 2. SCHEDULE a meeting with the stakeholder the week BEFORE the board meeting to go through metrics needed for the meeting. 3. LEARN what is IMPORTANT to the success of the business, not a person and build around that. 4. Be up front with the stakeholder about what expectations will be missed if you have to do urgent requests. 5. Track urgent requests so you can report these metrics to the business and its impact on productivity. Nothing is perfect, but these will help. #data #analytics #rockstar

13

2 Comments

😱 I once made a really bad dashboard on purpose.   🏆 Then someone challenged me to explain how to fix it.   🔨 That was a lot of extra work, but I did it.   🖱️ I loaded it up with show/hide text buttons with explanations and links to external resources.   🚛 It's a truckload of data viz design knowledge in one place.   🔧 The dashboard has been adjusted after each of several presentations to university students in the U.S., Canada, and the U.K.   🏈 This is a Throwback Thursday viz!   🏗→🧠 Build to Learn!   💭🚶‍♀️🚶‍♂️ Follow for more.   #tableau #data #analytics #VizoftheRay

12

3 Comments

Tried to take a different approach to #MakeoverMonday this week. I used some techniques from #WorkoutWednesday I learned a couple weeks ago for handling colors. With Tableau especially I have found that learning seemingly small tricks over time is a great way to upskill and create better visuals. https://lnkd.in/gnjuT7QT #dataviz #datafam

9

Building a Strong Data Team: My Insights from 10+ Years in the Data Field! 📣 For over a decade, I've been elbow-deep in the exciting world of data. And one thing I've learned for sure is that building a successful data team isn't about having everyone be the same kind of expert. The best teams win when everyone brings a different skill to the table, creating a well-rounded crew. Here's a breakdown of the key areas I've found to be crucial, based on my real-world experience. I have also provided course link in comments. 1️⃣ FOUNDATION SKILLS FOR EVERYONE: ✅ Python Programming: Think of Python as the common language many data folks use to work with data. It's a great skill to have, and there are fantastic resources to get you started, no matter your experience level. Even if you don't become a master coder, having a basic understanding is a huge plus. ✅ Data Warehousing Basics: This might sound fancy, but it's really just understanding how data is organized and stored. ✅ Product Analytics(Optional): If you want to understand how people interact with a product (website, app, etc.), this is your jam. It will teach you how to track user behavior and get valuable insights. ✅ Understanding Paid Ads (Optional): Data and marketing teams often work together, and knowing how paid advertising works can bridge the gap. It can help you understand the data coming from those campaigns. 2️⃣ BUILDING & MANAGING DATA SYSTEMS: ✅ Version Control: Imagine a "save point" for your data projects, like in a video game. Version control systems like Git and GitHub let you track changes and revert to earlier versions if needed. ✅ Automation: Let's face it, some data tasks can be repetitive. Automation tools like Jenkins/Airflow can help streamline those processes, freeing you up for more creative work. 3️⃣ DATA SCIENCE ESSENTIALS: ✅ Statistics for Data Science: Data is all about numbers, so understanding some basic statistics is key. ✅ Machine Learning 101: This is a powerful area of data science, but don't worry, you don't need a PhD to grasp the core concepts. The resources in comments will give you a solid foundation. 4️⃣ DEEP DIVES FOR DATA SCIENTISTS: If you're a data scientist who wants to specialize in a particular area, like machine learning or deep learning, there are tons of fantastic courses available. ✅ Supervised ML, Unsupervised Deep Learning ✅ Support Vector Machines, Hidden Markov Models ✅ Cluster Analysis ✅ Understanding Generative AI with LLMs The beauty of a strong data team is that everyone doesn't need to be an expert in everything. The true power comes from having a diverse group of people with complementary skills. Keep learning, stay curious, and be awesome!  For more insights follow me here - https://lnkd.in/gK2U8KtJ #LIPostingChallengeIndia #data #analytics #datajourney #datascience

19

3 Comments

'Data isn't easy. It's incredibly complex.' Yes, I'm just stealing Dylan Anderson's newsletter headlines for my posts these days. And data is complex. Yet, as with everything we overcomplicate it massively. And we do so in a variety of ways. Processes We have been doing 'data' for decades, and technology helps us scale and do things faster than we used to. But this means we now attempt to do everything. If you previously had to wait a day for your query to run, you made damn well sure it was worth running. Tooling How many ways can you complete a simple data task? And in how many different tools? Yet the fundamentals are still the same. Now we have to join a load of disparate systems together. But you can get another tool which can do that... Job Roles Anyone know where to start with this one? Dylan makes the point that in order to deal with such complexity we develop silos to specialise in each area. This makes the problem worse as communication lines fail. Data teams end up siloed and further away from the areas in which they can have a business impact. And then we overcomplicate things as this gap widens. Some processes require complexity. Most don't. Where are you making your work more complicated than it needs to be? Discover more of my posts by following hashtag #datatranslator #data #analytics #technology P.S. check out Dylan's newsletter, it will get you thinking

17

2 Comments