JP Morgan Interview Questions | Underfitting
Question
A model experiencing Underfitting has a poor predictive performance. Suggest a method to ensure that the model is free from Underfitting.
in progress
0
Machine Learning
4 years
4 Answers
1026 views
Great Grand Master 0
Answers ( 4 )
we can include more features or use a high dimensional algorithm like svm or even neural networks
Underfitting is a characteristic of a model, it means your data is not fitting well with your data.
Technically, the model is of lower degree.
To ensure the model is free of underfitting, either increase your order of your model or increasing the parameters of your ML model
Some ways to tackle under fitting are
1) Increase the no of parametres in the model(can make use of domain knowledge here).
2) Increase training data
3) Increase the duration of training to get better results.
4) Perform Feature engineering.
Getting more training data may not help as the model itself isn’t understanding the pattern within the already present data, Adding more data in context of additional rows won’t help as it isn’t learning the pattern.