Can we achieve better accuracy that SKLearn built-in functions?
Answer is Yes. In this notebook, I will demonstrate how Sigmoid Neurons implemented from scratch can provide better accuracy than SKLearn built-in Logistc Regression funtion.
SKLearn Logistic Regression implements Sigmoid curve for classification. However, it is more generalised. A Sigmoid Neuron implemented from scratch gives better accuracy(5-7% better accuracy) over SKLearn LogisticRegresstion funtion.
Accuracy Achieved:
- SKLearn LogisticRegression: 89.47368421052632 %
- Sigmoid Neurons: 94.73684210526315 %
Improvement in accuracy of Sigmoid Neurons over SKLearn LogisticRegression = 5.263157894736835 %