Personal Skills Development — Marketing Data Scientist
We work from home, we socialize online and we go to the movies in our living rooms. Most of society has truly learned to embrace a digital life. It wasn’t as much of a change for me as I’ve been working remotely for over 10 yrs.
When you are given lemons, make lemonade. As I noted in a previous blog post… there are alot of resources available online. I thought I’d share some lessons I learned through my personal development journey through a tumultuous year.
I first became interested in data science about 7–8 yrs ago while looking for a way to improve our lead management processes which lead me to a number of predictive lead scoring vendors. We had already had rules-based lead scoring for a number of years and this seemed like the natural next evolution and while I understood the general approach I didn’t have a grasp of the nuances of the different modeling techniques and the trade-offs between them. I would gain some gems of knowledge here and there for the next few years through projects and interactions with Data Science teams but these were random and progress was slow.
I believe(d) we are at the beginning of a massive skills shift. Just as years ago, marketers were expected to become more analytical and examine the performance of their campaigns, marketers will soon need to add data science skills to their toolset as application begin to add machine learning. 2020 brought turmoil to the job market and so it seemed like a good time for a career break to add the skills I’ve been admiring for so long.
Though most of the learning resources are still programmer oriented I did find some gems aimed at the technical business person. And so starts my journey toward becoming a citizen marketing data scientist began. As I stumbled through this confusing journey I will come to the realization that prioritizing in a hierarchy the resources (once found) would be a more efficient way to learn:
- Nano Degree
- Misc (Videos, Articles, Blogs, etc)
Benefit — This is a great place to start (or start with an introductory course to settle on a nano degree) The nano degree covers a wide number of topics and as the name suggests reminded me of my University education… start with a wide set of fundamental topics to help you see the landscape and so you can decide on what you want to specialize your learning to.
Commitment — 4–6 months depending on the amount of time you can put in per week
Structure — Multiple courses, each with projects culminating in a large capstone project that combine concepts across courses, access to mentors for help and projects are returned with human review and comments and need to be resubmitted until completed properly
Taken in 2020 — Udacity: School of Data Science: Predictive Analytics for Business. I loved the way this program is structured and supported by real people. You started with a couple of hours of lessons (pre-recorded videos) for each course. There are automated quizzes at checkpoints to test your understanding. Sometime partway through there are projects but definitely, at the end of each course, there is a project that uses the concepts taught. This is where I loved having access to peer groups to ask simple questions (I could get the app to do this or how did you configure that?) and access to a mentor to ask deeper questions about the material or project. While this isn’t real-time I found I usually get a response within an hour. The projects feel like real-world problems I’ve seen during my career so they feel relevant. And just like the real working world, you keep working on the project until what you submit is what is exactly what is expected. This approach is wonderful compared to University Engineering (as I remember it) because you keep learning and improving rather than get a mark and move on without learning what you missed. Because these are practical applications vs tests in a fixed time frame… the scope is larger than can be covered in an automated test and is marked and commented on by a real human. In this nano degree, the courses definitely built on top of each other and the final capstone project combined concepts from all the courses. I feel the project approach separates the people that can do the work vs those that can only regurgitate definitions and talk a good game. Can’t say enough good things about the program. If you are interested these are the topics covers in the courses of this nano degree:
- Master a framework for solving complex problems with data
- Identify the most appropriate analytical methodology for a business problem
- Clean, format, and blend data in minutes vs hours in Excel
- Implement a variety of predictive modeling techniques, including linear regression, classification, A/B — test, forecasting, and clustering models
- Visualize and communicate data and insights effectively
Benefit — Having seen Data Science at the high level (theory) you’ll need to refine your skillset on hands-on tactical application of Data Science. Many of these certifications tended to be platform focused and the content was more about how to configure/code on that platform. In the past, I have taken certifications that are more about the theory and process vs tactical how-to instruction. Each certification definitely felt more narrow in scope as compared to a nano degree. They are also a good way to show prospective employees and teammates what skills they can expect you to be proficient in.
Commitment — There was quite a lot of variation here 1–8 week seemed like the typical duration
Structure — These typically had multiple courses but not as many hours in classroom time as a nano degree. There are optional practice assignments that are not marked and getting the certification depended on getting a passing grade (most seem to have settle for 80%). Generally, I don’t like the timed automated testing in the certifications I have taken recently and in the past as I feel they are only able to test superficial understanding and definitions.
Taken in 2020 — In the past I’ve been certified for various Marketing processes and platforms but last year I wanted to focused on Data Science.
- Alteryx Designer Core — As the name suggests this focuses on the Alteryx Designer platform. The material is structure into an 8-week program with material organized in days to help set a pace. There are hours of videos that introduce concepts followed by practice exercises after major milestones. These aren’t marked but includes a solution for you to compare your work with. Topics included data import, clean, standardize, and transform values and layout for data analysis and modeling. These culminated with a Certification Exam. This was one of the best certification exams I have taken. The testing was intense (reminded me of engineering exams)… 2-hour time limit for 80 questions and you will need most of the time. I must admit the first time through I took too much time trying to be perfect and ran out of time. Some of it was definitions but there were a number of practical application questions that meant you need to download the problem set and work in the app to process the data to get to a set of answers. If you don’t pass the first time you are given a break down of subject areas and how you scored which gives you a road map of where to focus your study time
Learn more about the Alteryx Academy and Certification
- Data Science with Knime Software (Level 1) — This certification was mostly about being proficiency in the Knime Workbench platform, preparing and formating data, data analysis, and setting up basic models to learn the patterns and predict outcomes. There are 4 groups of lessons each with video content and supplemental reading that prepare you for a practice exercise where you can compare your solution with a supplied solution. The certification exam is time-limited per question and runs about 45 mins.
Learn more about KNIME Analytics Platform for Data Scientists: Basics
- Data Science with Knime Software (Level 2) — Building on the basics, this certification covers processing Time Series Data, connecting and working with database calls, grouping functions for automated processing, abstracting your processes to be reusable for other workflows, and advanced machine learning. The lessons follow the same format as the basic level with videos and supplemental reading to prepare you for the practice exercises that have solutions for you to reference. Just as before the certification exam is time-limited per question and runs about 45 mins.
Learn more about KNIME Analytics Platform for Data Scientists: Advanced
Benefit — These are a great way to get familiar with a subject area and cover a similar amount of detail as one of the courses/lessons that are part of a certification. If you are not sure if you want to commit to a nano degree, take an introductory course in Data Science to get an understanding of the landscape of the major subject areas and to start to see what you might want to dive deeper into. No exam at the end. Sometimes these courses are the same courses that are part of certifications and nano degrees. Books can also fall into this category.
Commitment — 1–10 day to complete (no exam)
Structure — A set of pre-recorded videos broken up into lesson by category. Sometimes has mini quizzes and practice exercises.
Taken in 2020 — I focused on skills I thought would help me build a tech stack to process data, customize models, communicate insights, and have users interact with the data.
- KNIME: Time Series Analysis
- Udacity: Tableau Data Visualization
- Pluralsight: Introduction to SQL
- Pluralsight: Core Python: Getting Started
- Pluralsight: Getting started with Programming in Java
- Pluralsight: Getting started with Linux
Misc Videos, Articles, and Blogs
Benefit — Where I find these to me most beneficial is after having a good foundation (taking a nano degree or a course) and just need to see the mechanics of how something needs to be setup
Commitment — the commitment varies a lot but generally take 10–45 min to consume
Structure — varies based on the medium
Taken in 2020 — Various… too many to include and usefulness is very situational
I stumble through this learning journey and didn’t start from a broad view of the landscape of Data Science and narrow it down to more specific skills. If I were to go back in time I’d definitely use the hierarchical approach I outlined.
For future learning, my plan is to expand my understanding in 2 directions.
- Expand Understanding of Statistics
- Intro to Descriptive Statistics
- Intro to Inferential Statistics
2. Execution — new tools and techniques
- KNIME: Server — Productionizing and Collaboration
- Implementing Hidden Markov Models in Python
- Learn about the Weka Workbench
I’ll be spending more time with this resource, Machine Learning Mastery. There is so much to learn here but it is geared to developers.
Comment on my blog about your learning journey last year and what you want to learn this year. Let me know if there are Marketing Data Science topics you would like to see talked about.
Originally published at https://marketingoptimized.wixsite.com on January 4, 2021.