ML/AI

Question 1

Supervised vs Unsupervised vs Reinforcement Learning?

Answer 1

In SL, the model learns form the labeled data and makes a prediction as the output.
In USL, the model uses unlabeled input data and allows the algorithm to act on that information without guidance
In RL, the model learns based on the rewards it received for its previous actions.

Question 2

What is overfitting?
Ways to avoid?

Answer 2

Overfitting is a situation, where a model learns the training set too well which causes random fluctuations to be considered as concepts in training data.
1. Regularization
2. Simpler model, less variables/parameters
3. Cross-validation(k-folds)
4. LASSO

Question 3

What is Bias and Variance in ML model?

Answer 3

Bias is a machine learning model and occurs when the predicted values are further from the actual values. Low bias means good(closer to value).
High bias can cause underfitting, miss relevant relations.

Variance is the amount of target model will change when trained with different training data.
Variance should be minimalized.
High variance can cause algo to model the random noise in the training data instead of expected output.

Question 4

How do you handle missing or corrupted data?

Answer 4

IsNull() , dropna(), fillna()

Question 5

How Can You Choose a Classifier Based on a Training Set Data Size?

Answer 5

When training data is small, a model that has right bias and low variance seems to work better.
e.g., Naive Bayes works best with large data set

Question 6

What Are the Three Stages of Building a Model in Machine Learning?

Answer 6

Model Building
Model Testing
Applying the Model

Question 7

Deep Learning?

Answer 7

The Deep learning is a subset of machine learning that involves systems that think and learn like humans using artificial neural networks.

Question 8

Difference b/w ML and DL?

Answer 8

Feature engineering is done manually in ML whereas in DL, it consists of neural network which automatically determine which features to use.