Machine Leaning Interview Questions


Govind Choudhary Nov 11 2021 · 2 min read
Share this

1.    How you can define Machine Learning?

Ans. We can define Machine Learning as an application of artificial intelligence where we humans try to find relationships among the data and then try to feed that relationships about the data to machine or computer so that it will be able to predict the data in future based on the relationships we have fed to the machine or computer without being explicitly programmed.

2.    What do you understand Labelled training dataset?

Ans. Basically a labelled training dataset is a dataset where we have both features as well as target, it means we have the input as well as output, here input means the data on basis of which we want to predict the output and output is our predicted data (which in case of labelled training set is already known.) so labelled training dataset has both.

3.       What are 2 most common supervised ML tasks you have performed so far?

Ans. The 2 most common supervised ML tasks I have performed so far is Linear Regression and Decision Tree.

4.    What kind of Machine learning algorithm would you used to walk robot in various unknown area?

Ans.  I would like to use Reinforcement Machine Learning algorithm where it can learn from responses and optimize itself.

5.       What kind of ML algo you can use to segment your user into multiple groups?

Ans.   We can use unsupervised machine learning algorithms or clustering techniques to segment  our user into multiple groups more specifically I would like to go with DBSCAN clustering     technique as it is fast and gives best result when compared with other clustering techniques.

Example of DBSCAN algorithm

6.       What type of learning algo realised on similarity measure to make a prediction?

Ans. The Learning algorithm that realised on similarity measure to make a prediction is instance based learning algorithm, example is K Nearest Neighbours(KNN) algorithm.

7.       What is an online learning system?

Ans. An online learning is an approach that ingests data one observation at a time, in online learning once the data or observation is consumed it is no more required, i.e once the model is deployed into production if the new data comes so it will train the model on the way only for that data or observation it does not need to retrain it again with complete data as we do in offline learning.

Basic architecture of online Learning system

8.       What is out of core learning?

Ans.  Out of core learning simply means that when a computer is unable to fit the data with any certain algorithms on memory of a single computer. In more simple words, Out of core learning is working on data that is too big to fit on ram.

9.       Can you name couple of ml challenges that you have faced?

Ans. i) Less Amount of Training Data

         ii) Poor Quality of Data

        iii) Ovetfittting of the model

        iv) Underfitting of the model

        v) Non-representative Data

        vi) Imbalanced Dataset

10.   Can you please give 1 example of hyperparameter tuning wrt some classification algorithm?


           grid = dict(n_estimators=n_estimators,max_features=max_features)

          GridSearchCV(estimator=RandomforestClassifier(), param_grid=grid, n_jobs=-1,   cv=5,scoring='accuracy',error_score=0)

Read next