Hello :) Today is Day 308!
A quick summary of today:
- covered module 4 - linear models
- covered module 5 - decision tree models
Sklearn MOOC module 4 - Linear models
Main takeaways:
- the predictions of a linear model depend on a weighted sum of the values of the input features added to an intercept parameter
- fitting a linear model consists in adjusting both the weight coefficients and the intercept to minimize the prediction errors on the training set
- to train linear models successfully it is often required to scale the input features approximately to the same dynamic range
- regularization can be used to reduce over-fitting: weight coefficients are constrained to stay small when fitting
- the regularization hyperparameter needs to be fine-tuned by cross-validation for each new machine learning problem and dataset
- linear models can be used on problems where the target variable is not linearly related to the input features but this requires extra feature engineering work to transform the data in order to avoid under-fitting
I streamed the whole module, however I was in the library so I could not turn on the mic/cam.
Part 1 and after a brief internet problem I recorder Part 2
Using the hosted env (through binder) is so nice because there is no copilot and if I want to write code I have to find the functions/imports/params myself which again makes me learn about functions’ params and creates a virtuous cycle of learning. Copilot can really have a bad influence when learning/reviewing for a coding exam 😆
Sklearn MOOC module 5 - Decision tree models
Main takeaways, they are:
- are suited for both regression and classification problems
- are non-parametric models
- are not able to extrapolate
- are sensitive to hyperparameter tuning
Very nice module again. The stream was 2 hours ~ Here is a link.
I am really enjoying this MOOC and is a must-do course for a beginner that wants to learn about sklearn and some of the most popular non-DL models. And for sure in the future if I have to teach some of the material, I can refer people to this or even refer to it when I’m preparing my materials.
Overall, I streamed for 5 hours today which is amazing, and also makes me a bit exhausted 😆
That is all for today!
See you tomorrow :)