Mistakes to Avoid in Machine Learning
Mistakes to Avoid in Machine Learning
What you’ll learn
Skills you’ll gain
It's exciting to build machine learning models, but dealing with errors, bad output or a host of other issues can slow your progress. In this fast-paced course from Patagonia data scientist Brett Vanderblock, you'll learn the mistakes you should avoid when building machine learning models. you'll explore how to better work with experts, standardize your data, prevent data leakage, and how to give better presentations. From dealing with bad data, to preventing overfitting, to not getting feedback, you'll know how to avoid key mistakes when building your machine learning models.
Syllabus
Download syllabus-
1
Assuming data is good to go Assumptions can spell disaster. After watching this video you'll be able to determine if your data is actually ready for modeling. 2m
-
2
Neglecting to consult subject matter experts Make sure you are solving the question that's being asked. 1m
-
3
Overfitting your models After watching this video you'll be able to prevent overfitting in your predictive models. 2m
-
4
Not standardizing your data Standardizing is a step for a reason, so don't skip it! After watching this video you'll be able to normalize your data effectively. 2m
-
5
Focusing on the wrong factors Don't spend time perfecting your models if your data isn't adequate. 2m
-
6
Data leakage It's possible when working with data that the wrong data is included in your model. After watching this video you'll be able to prevent data leakage. 2m
-
7
Forgetting traditional statistics tools If you need to understand the past, traditional regression techniques might be a better option. 1m
-
8
Assuming deployment will be a breeze Don't get caught spending hours fixing your model at the last minute. 1m
-
9
Assuming Machine Learning is the answer Sometimes Machine Learning is not the best method. 1m
-
10
Developing in a silo Your work is better when you collaborate. After watching this video you'll be able to seek valuable input from others regarding your code. 2m
-
11
Not treating for imbalanced sampling A common error in Machine Learning is having a disproportionate ratio of observations in each class you're working with. 3m
-
12
Interpreting your coefficients without properly treating for multicollinearity After watching this video you'll be able to treat your models for multicollinearity. 3m
-
13
Evaluating by accuracy alone After watching this video you'll be able to evaluate your models effectively using a variety of checks and balances. 6m
-
14
Giving overly technical presentations Avoid making a confusing pitch with too much technical jargon. 1m
-
1
Take your Machine Learning skills to the next level 1m
Certificate
Certificate of Completion
Awarded upon successful completion of the course.
Instructor
Brett Vanderblock
Brett Vanderblock is a data scientist with Patagonia and co-founder of Think Fast Analytics. He has proven experience surfacing insights across industries, from start-ups, state and local governments, higher education, healthcare, to retail. He is always furthering his deep expertise in machine learning, data visualization, and data pipelines. An educator at heart, Brett enjoys providing "data therapy" to enable others to realize the benefits of their data through self-service analytics.
Brett Vanderblock
Data Scientist
Accreditations
Link to awards
How GoSkills helped Chris
I got the promotion largely because of the skills I could develop, thanks to the GoSkills courses I took. I set aside at least 30 minutes daily to invest in myself and my professional growth. Seeing how much this has helped me become a more efficient employee is a big motivation.
