A company has build a kNN classifier that gets 100% accuracy on training data. When they deployed this model on client side it has been found that the model is not at all accurate. Which of the following thing might gone wrong? Note: Model has successfully deployed and no technical issues are found at client side except the model performance

it is probably a overfitted model
it is probably a underfitted model
can't say
wrong client data

The correct answer is A. It is probably an overfitted model.

Overfitting occurs when a model learns the training data too well and is unable to generalize to new data. This can happen when the model is trained on a small amount of data or when the data is not representative of the real world.

In this case, the company has built a kNN classifier that gets 100% accuracy on training data. This suggests that the model has learned the training data very well. However, when the model is deployed on the client side, it is found that the model is not at all accurate. This suggests that the model has overfitted the training data and is unable to generalize to new data.

There are a few things that the company can do to prevent overfitting. One option is to increase the amount of training data. This will help the model to learn the patterns in the data more accurately. Another option is to use a different machine learning algorithm that is less likely to overfit.

Here is a brief explanation of each option:

  • Option A: It is probably an overfitted model. This is the most likely explanation for the problem. The model has learned the training data too well and is unable to generalize to new data.
  • Option B: It is probably an underfitted model. This is less likely, but it is possible that the model has not learned the training data well enough. This can happen when the model is trained on a large amount of data or when the data is very noisy.
  • Option C: Can’t say. This is not a good option, as it does not provide any useful information.
  • Option D: Wrong client data. This is also not a good option, as it is unlikely that the client data is wrong.

I hope this helps! Let me know if you have any other questions.