Learn everything about bias and variance in machine learning

What is Bias in machine learning?

In ml model, there is an actual value in the dependent variable and the machine predicts a value according to the data.
So bias is the distance/gap/difference between the actual value and predicted value. High bias means the gap is very much and low bias means the distance or gap is less or can be no gap or distance. Low bias model makes fewer assumptions about the form of the target function. A high bias makes more assumptions and for this reason it's unable to capture the important features of the dataset.

Ways to reduce high Bias:
1. Increase the input features if the model is underfitted.
2. Decrease the regularization term.

What is Variance machine learning?

Variance means all the predicted value which was predicted by the model, how much scatter they are in relation with each other. Low variance means all the predicted values are in a group mean the values will not be so much scattered or the predicted values will be near from each other. High variance means all the predicted values are not in a group means all the values will be so much scattered or the predicted values will be far from each other. If the model variance is high then it will leads to overfitting problem.

Low-Bias, Low-Variance:
Low bias and low variance combination is a very good model we can and ideal machine learning model. But the problem is practically this model is not possible.
Low-Bias, High-Variance:
combination of low bias and high variance model, gives on average accurate predictions . This type scenario occurs when we train a model using a large number of parameters. t leads to overfitting problems in the model
High-Bias, Low-Variance:
Combination of High bias and low variance, gives consistent predictions but inaccurate on average. This type of scenario occurs when our model is not well trained with the training dataset. It leads our model to a underfitting problem.
High-Bias, High-Variance:
Combination of high bias and high variance, a model gives inconsistent and inaccurate predictions on average.

How to identify High variance or High Bias?

High bias and variance can be identify:
1. if the training error and high test error.
2. If the high training and test error is almost similar to training error.