Langit
LightNews LightNews
glangit.bsky.social
Langit
@glangit.bsky.social
3 followers 8 following 2 posts
A machine learning enthusiast.
Posts Replies Media Videos
glangit.bsky.social
First year in data science. And also my first logistic regression, just sklearn and the parameters. So, i created the second model and adding more parameter than before. This is what i got, i sacrificed the accuracy actually, but this is inline with my goals (lowering the False-Negative).
First year in data science. And also my first logistic regression, no fine-tune, no tree-based, just sklearn and the parameters. I use the aussie rain data (time series), and my goal is to reduce the False-Negative rate so that the model does not assume “NO” too often for the “RainTomorrow” feature. I created 2 models, the first one is actually high in accuracy (85%) and low in False-Positive, but the False-Negative was too high, which is not in line with my goal. So, i created the second model and adding more parameter than before. This is what i got, i sacrificed the accuracy actually, but this is inline with my goals (lowering the FN). And i use the model to a new data, and the result is fine (the FN was 0.25 or 25%).
January 11, 2026 at 4:25 PM