Progress Report: Day 53-54
Current stage: calc & coding Learned for: 6h yesterday 6h today Satisfied? yes
made two very basic linear regression models for mapping the relationship between hours studied & test scores for students. (details below)
day 1:
model 1: created synthetic data and played around with various Numpy rand functions model 2: downloaded a random Kaggle dataset to practice on real data
my pseudo-code: 0. create a synthetic dataset or download a real one
- use pandas to manipulate the file & wrap it into a data frame
- split the data for training and validation
- predict on the data using scikit-learn lib
- evaluate (R2, MSE)
- plot a regression line on a graph using Matplotlib (screenshots below)
day 2:
watched the Beginner Intro to Neural Networks video series by giant_neural_network on YouTube.
I really liked how he visualized the use of derivatives from calculus to minimize the loss function regardless of the starting point on the curve.
I felt happy finally understanding the basic math of it. Unfortunately, I still haven’t learned partial derivatives, so I didn’t memorize the formula when there are multiple variables. RIP.
Overall happy with the progress. 🫡
Screenshots of model 1 (synthetic data) plot with regression line below.
Some stats: Intercept: 19.94 Coefficient for Hours Studied: 8.05 R² score: 0.95 Mean Squared Error: 28.22 Predicted Test Score for 4 hours of study: 52.14
Model 2 (Kaggle data):
Intercept: 75.34 Coefficient for Hours Studied: 0.45 R² score: 0.09 Mean Squared Error: 132.26 Predicted Test Score for 10 hours of study: 79.79
Needless to say, these students have a shit ROI on any additional learning outside these classes 😂.
What I’ve done well last 2 days:
- Started fixing sleep schedule, now <1:30 AM
- Improved diet by a bit
- Went to gym (chest & legs)
- Made more friends & am fired up to keep improving
Bad: Not really bad, but I also need to catch up on client work that is slowly piling up.