| # | Tweet | Community | Topic | Views ▼ | Ratio | Engagement | Posted |
|---|---|---|---|---|---|---|---|
| 1 | [image] Day 3 of building ML models - Decision Trees & Random Forests.
Predicted loan approval from 491 applications. Random Forest edged out a single Decision Tree (84.85% vs 83.84%), no overfitting on either. | Machine Learning- Data Science | — | 456 | 0.2x | 12 | Sep 11 |
| 2 | [image] Day 2 of building ML models - Logistic Regression
Built a Logistic Regression churn model, 80% accuracy sounded great until I checked recall and found it was missing half of the actual churners.
Fixed it by tuning the decision threshold, not the model | Machine Learning- Data Science | — | 358 | 0.2x | 12 | Sep 10 |
| 3 | [image] Day 6 of building ML models one at a time: Naive Bayes.
Built a spam classifier on 5K+ SMS messages. 98.4% accuracy AND 91% recall on spam, no trade-off this time, unlike my churn and wine models.
Full project: | Machine Learning- Data Science | — | 348 | 0.2x | 10 | Sep 17 |
| 4 | [image] Spent the weekend turning 55k messy hospital records into a clean power BI dashboard
Data cleaning - DAX - Custom theme, start to finish📊 | Data Analytics Africa | — | 307 | 0.1x | 14 | Sep 7 |
| 5 | [image] Day 4 of building ML models - K-Means Clustering
Used K-Means to turn 200 anonymous mail custominto 5 marketing personas, no labels, just clustering by income and spending behavior.
Full project: | Machine Learning- Data Science | — | 288 | 0.1x | 8 | Sep 15 |
