Today's Interview QnAs
Company: Celstream Technologies
Role: Data Scientist
1. Why is KNN algorithm called Lazy Learner?
Ans: When it gets the training data, it does not learn and make a model, it just stores the data. It does not derive any discriminative function from the training data. It uses the training data when it actually needs to do some prediction. So, KNN does not immediately learn a model, but delays the learning, that is why it is called lazy learner.
2. What is stratify in Train_test_split?
Ans: Stratification means that the train_test_split method returns training and test subsets that have the same proportions of class labels as the input dataset. So if my input data has 60% 0's and 40% 1's as my class label, then my train and test dataset will also have similar proportions.
3. Explain the difference between lists and tuples.
Ans: Both lists and tuples are made up of elements, which are values of any Python data type.
Lists are mutable, while tuples are immutable.
Lists are created using Square brackets (for example, my_list = [a, b, c]), while tuples are in parentheses (for example, my_tuple = (a, b, c)).
4. What is the difference between classification and regression?
Ans: Classification is about predicting a label and regression is about predicting a quantity. In classification we predict a discrete class label . Examples of Classification algorithm are Decision Tree, KNN, Naive Bayes, Logistic regression etc.
In regression we predict a continuous quantity output. Examples of regression models are Linear regression, Random Forest regressor etc.
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
Company: Celstream Technologies
Role: Data Scientist
1. Why is KNN algorithm called Lazy Learner?
Ans: When it gets the training data, it does not learn and make a model, it just stores the data. It does not derive any discriminative function from the training data. It uses the training data when it actually needs to do some prediction. So, KNN does not immediately learn a model, but delays the learning, that is why it is called lazy learner.
2. What is stratify in Train_test_split?
Ans: Stratification means that the train_test_split method returns training and test subsets that have the same proportions of class labels as the input dataset. So if my input data has 60% 0's and 40% 1's as my class label, then my train and test dataset will also have similar proportions.
3. Explain the difference between lists and tuples.
Ans: Both lists and tuples are made up of elements, which are values of any Python data type.
Lists are mutable, while tuples are immutable.
Lists are created using Square brackets (for example, my_list = [a, b, c]), while tuples are in parentheses (for example, my_tuple = (a, b, c)).
4. What is the difference between classification and regression?
Ans: Classification is about predicting a label and regression is about predicting a quantity. In classification we predict a discrete class label . Examples of Classification algorithm are Decision Tree, KNN, Naive Bayes, Logistic regression etc.
In regression we predict a continuous quantity output. Examples of regression models are Linear regression, Random Forest regressor etc.
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
Today's Interview QnAs
Company: TCS
Role: Data Scientist
1. What is Information gain?
Ans: Information gain is the difference of entropy of the parent node and the weighted entropies of the child nodes.
Information Gain = Entropy before splitting - Entropy after splitting
Suppose we have multi features in a dataset, so we will choose that feature as a root node which will have maximum Information Gain.
Information gain is used for determining the best features/attributes that render maximum information about a class. It follows the concept of entropy while aiming at decreasing the level of entropy, beginning from the root node to the leaf nodes.
2. When accuracy should not be used as a parameter to measure the performance of Classification model?
Ans: There are 2 major situations when you will not use accuracy as a parameter for classification performance measures-
A) When you have severely imbalanced dataset. If suppose you have a dataset with 90% as positive class and 10% as negative class, then even a dumb model can classify all the datapoints as positive class and attain the accuracy as 90%, which is not sensible.
B) Since accuracy does not considers probability values so if I have a threshold of 0.5 and my model M1 gives the probability value for a "x" as 0.9 and my second model gives the probability value of 0.55, both of them will be labelled as positive class but we know the performance of M1 is much better than M2 when you see the actual probabilities.
3. Why do we use the summary function?
Ans: Summary functions are used to give the summary of all the numeric values in a dataframe. Eg. The describe() function can be used to provide the summary of all the data values given to it.
column_name.describe() will give the following values of all the numeric data in the column-
Count
Mean
Std-Standard deviation
Min-Minimum
25%
50%
75%
max-Maximum
4. How are Entropy and Gini index related?
Ans: Both Entropy and Gini index are parabolic in nature when it's plotted by the probability. However the maximum value of my Gini Impurity is 0.5 while that of entropy is 1 which only occurs when my classes are Equi-probable. Both of them will tend to zero values when one of the classes are dominant.
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
Company: TCS
Role: Data Scientist
1. What is Information gain?
Ans: Information gain is the difference of entropy of the parent node and the weighted entropies of the child nodes.
Information Gain = Entropy before splitting - Entropy after splitting
Suppose we have multi features in a dataset, so we will choose that feature as a root node which will have maximum Information Gain.
Information gain is used for determining the best features/attributes that render maximum information about a class. It follows the concept of entropy while aiming at decreasing the level of entropy, beginning from the root node to the leaf nodes.
2. When accuracy should not be used as a parameter to measure the performance of Classification model?
Ans: There are 2 major situations when you will not use accuracy as a parameter for classification performance measures-
A) When you have severely imbalanced dataset. If suppose you have a dataset with 90% as positive class and 10% as negative class, then even a dumb model can classify all the datapoints as positive class and attain the accuracy as 90%, which is not sensible.
B) Since accuracy does not considers probability values so if I have a threshold of 0.5 and my model M1 gives the probability value for a "x" as 0.9 and my second model gives the probability value of 0.55, both of them will be labelled as positive class but we know the performance of M1 is much better than M2 when you see the actual probabilities.
3. Why do we use the summary function?
Ans: Summary functions are used to give the summary of all the numeric values in a dataframe. Eg. The describe() function can be used to provide the summary of all the data values given to it.
column_name.describe() will give the following values of all the numeric data in the column-
Count
Mean
Std-Standard deviation
Min-Minimum
25%
50%
75%
max-Maximum
4. How are Entropy and Gini index related?
Ans: Both Entropy and Gini index are parabolic in nature when it's plotted by the probability. However the maximum value of my Gini Impurity is 0.5 while that of entropy is 1 which only occurs when my classes are Equi-probable. Both of them will tend to zero values when one of the classes are dominant.
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
❤1
Today's Interview QnAs
Company: Splunk
Role: Data Scientist
1. Explain ways to train hyperparameters in a neural network?
Ans:Hyperparameters in a neural network can be trained using four components:
1. Batch size: Indicates size of input data
2. Epochs: Denotes the number of times the training data is visible to neural network to train
3. Momentum: Used to get an idea of next steps that occur with data being executed
4. Learning rate: Represents time required for network to update parameters and learn
2. What happens if the learning rate is set too high or too low?
Ans: If the learning rate is too low, your model will train very slowly as minimal updates are made to the weights through each iteration. Thus, it would take many updates before reaching the minimum point.
If the learning rate is set too high, this causes undesirable divergent behavior to the loss function due to drastic updates in weights, and it may fail to converge.
3. Explain Crude working of LSTM? ==> How Does an LSTM Network Work?
Ans: Long-Short-Term Memory (LSTM) is a special kind of recurrent neural network RNN capable of learning long-term dependencies, remembering information for long periods as its default behaviour. Crude steps in an LSTM network are as:
1: Network decides what to forget and what to remember
2: It selectively updates cell state values
3: Network decides what part of current state makes it to output
4. Explain what is Swish Function?
Ans: Swish is an activation function proposed by Google which is an alternative to ReLU activation function.
It is represented as: f(x) = x * sigmoid(x)
Swish function works better than ReLU for a variety of deeper models.
The derivative of Swist can be written as: y’ = y + sigmoid(x) * (1 - y)
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
Company: Splunk
Role: Data Scientist
1. Explain ways to train hyperparameters in a neural network?
Ans:Hyperparameters in a neural network can be trained using four components:
1. Batch size: Indicates size of input data
2. Epochs: Denotes the number of times the training data is visible to neural network to train
3. Momentum: Used to get an idea of next steps that occur with data being executed
4. Learning rate: Represents time required for network to update parameters and learn
2. What happens if the learning rate is set too high or too low?
Ans: If the learning rate is too low, your model will train very slowly as minimal updates are made to the weights through each iteration. Thus, it would take many updates before reaching the minimum point.
If the learning rate is set too high, this causes undesirable divergent behavior to the loss function due to drastic updates in weights, and it may fail to converge.
3. Explain Crude working of LSTM? ==> How Does an LSTM Network Work?
Ans: Long-Short-Term Memory (LSTM) is a special kind of recurrent neural network RNN capable of learning long-term dependencies, remembering information for long periods as its default behaviour. Crude steps in an LSTM network are as:
1: Network decides what to forget and what to remember
2: It selectively updates cell state values
3: Network decides what part of current state makes it to output
4. Explain what is Swish Function?
Ans: Swish is an activation function proposed by Google which is an alternative to ReLU activation function.
It is represented as: f(x) = x * sigmoid(x)
Swish function works better than ReLU for a variety of deeper models.
The derivative of Swist can be written as: y’ = y + sigmoid(x) * (1 - y)
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
Today's Interview QnAs
Company: The Math Company
Role: Data Scientist
1. What are Different Kernels in SVM?
Ans: There are six types of kernels in SVM:
Linear kernel - used when data is linearly separable.
Polynomial kernel - When you have discrete data that has no natural notion of smoothness.
Radial basis kernel - Create a decision boundary able to do a much better job of separating two classes than the linear kernel.
Sigmoid kernel - used as an activation function for neural networks.
2. According to you what are some advantages in using a CNN over a DNN (dense neural network) for an image classification task?
Ans: Both models can capture relationship between close pixels but CNNs have some useful properties like: translation invariant — exact location of pixel is irrelevant for filter less likely to overfit — typical number of parameters in a CNN is much smaller than that of a DNN gives better understanding of model — one can look at filters, weights and visualize what network learned hierarchical nature — learns patterns in by describing complex patterns using simpler ones.
3. What are your intuitions about pruning in decision tree?
Ans: Pruning a decision tree helps to prevent overfitting the training data so that our model generalizes well to unseen data. Pruning a decision tree means to remove a subtree that is redundant and not a useful split and replace it with a leaf node. Decision tree pruning can be divided into two types: pre-pruning and post-pruning.
Pre-pruning, also known as Early Stopping Rule, is the method where the subtree construction is halted at a particular node after evaluation of some measure. These measures can be the Gini Impurity or the Information Gain. In pre-pruning, we evaluate the pruning condition based on the above measures at each node.
post-pruning means to prune after the tree is built. You grow the tree entirely using your decision tree algorithm and then you prune the subtrees in the tree in a bottom-up fashion. You start from the bottom decision node and, based on measures such as Gini Impurity or Information Gain, you decide whether to keep this decision node or replace it with a leaf node.
4. Give some intuition about Epoch, Batch, and Iteration in deep learning how they are different?
Ans:
1. Epoch - Represents one iteration over entire dataset (everything put into training model)
2. Batch - Refers to when we cannot pass entire dataset into neural network at once, so we divide dataset into several batches
3. Iteration - if we have 100,000 images as data and a batch size of 400. then an epoch should run 250 iterations (100,000 divided by 400)
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
Company: The Math Company
Role: Data Scientist
1. What are Different Kernels in SVM?
Ans: There are six types of kernels in SVM:
Linear kernel - used when data is linearly separable.
Polynomial kernel - When you have discrete data that has no natural notion of smoothness.
Radial basis kernel - Create a decision boundary able to do a much better job of separating two classes than the linear kernel.
Sigmoid kernel - used as an activation function for neural networks.
2. According to you what are some advantages in using a CNN over a DNN (dense neural network) for an image classification task?
Ans: Both models can capture relationship between close pixels but CNNs have some useful properties like: translation invariant — exact location of pixel is irrelevant for filter less likely to overfit — typical number of parameters in a CNN is much smaller than that of a DNN gives better understanding of model — one can look at filters, weights and visualize what network learned hierarchical nature — learns patterns in by describing complex patterns using simpler ones.
3. What are your intuitions about pruning in decision tree?
Ans: Pruning a decision tree helps to prevent overfitting the training data so that our model generalizes well to unseen data. Pruning a decision tree means to remove a subtree that is redundant and not a useful split and replace it with a leaf node. Decision tree pruning can be divided into two types: pre-pruning and post-pruning.
Pre-pruning, also known as Early Stopping Rule, is the method where the subtree construction is halted at a particular node after evaluation of some measure. These measures can be the Gini Impurity or the Information Gain. In pre-pruning, we evaluate the pruning condition based on the above measures at each node.
post-pruning means to prune after the tree is built. You grow the tree entirely using your decision tree algorithm and then you prune the subtrees in the tree in a bottom-up fashion. You start from the bottom decision node and, based on measures such as Gini Impurity or Information Gain, you decide whether to keep this decision node or replace it with a leaf node.
4. Give some intuition about Epoch, Batch, and Iteration in deep learning how they are different?
Ans:
1. Epoch - Represents one iteration over entire dataset (everything put into training model)
2. Batch - Refers to when we cannot pass entire dataset into neural network at once, so we divide dataset into several batches
3. Iteration - if we have 100,000 images as data and a batch size of 400. then an epoch should run 250 iterations (100,000 divided by 400)
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
Get complete hands-on practical learning experience of Data Science and become technically strong enough so you can crack any job interview.
Join Our Data Science Combo Course today.
Course Link- https://www.cloudyml.com/course-combo-package/
🔴NOTE: This offer gonna end today at 11:59PM. From tomorrow, you will not get this discount anymore so HURRY UP!
Join Our Data Science Combo Course today.
Course Link- https://www.cloudyml.com/course-combo-package/
🔴NOTE: This offer gonna end today at 11:59PM. From tomorrow, you will not get this discount anymore so HURRY UP!
🔴Job Openings🔴
Role: ML Developer
Company : Decimal Point Innovative Research Solutions
Exp - 0-2 yr
Link: https://ai.cloudyml.com/3ka5rvS
Role: Data Science Research Intern
Company: Prodigal
Exp - 0-1 yr
Link : https://ai.cloudyml.com/3LfJ43Q
Role: Product Analyst
Company: Samast Technologies Private Limited
Exp - 0-1 yr
Link: https://ai.cloudyml.com/3LeBPtb
Role: Data Analyst
Company: Enrich Talents
Exp - 0-3 yr
Link : https://ai.cloudyml.com/3vBlesS
Role: ML Developer
Company : Decimal Point Innovative Research Solutions
Exp - 0-2 yr
Link: https://ai.cloudyml.com/3ka5rvS
Role: Data Science Research Intern
Company: Prodigal
Exp - 0-1 yr
Link : https://ai.cloudyml.com/3LfJ43Q
Role: Product Analyst
Company: Samast Technologies Private Limited
Exp - 0-1 yr
Link: https://ai.cloudyml.com/3LeBPtb
Role: Data Analyst
Company: Enrich Talents
Exp - 0-3 yr
Link : https://ai.cloudyml.com/3vBlesS
Today's Interview QnAs
Company: Syngenta
Role: Data Scientist
1. What are Gaussian Mixture Model? Where they are used?
Probabilistic model for representing normally distributed subpopulations within an overall population. Mixture models in general don't require knowing which subpopulation a data point belongs to, allowing the model to learn the subpopulations automatically. Since subpopulation assignment is not known, this constitutes a form of unsupervised learning. These models have been used for feature extraction from speech data, and have also been used extensively in object tracking of multiple objects, where number of mixture components and their means predict object locations at each frame in a video sequence. change in place of 1st.
2. What are Hard-Margin and Soft-Margin SVMs?
Hard-Margin SVMs have linearly separable training data. No data points are allowed in the margin areas. This type of linear classification is known as Hard margin classification.
Soft-Margin SVMs have training data that are not linearly separable. Margin violation means choosing a hyperplane, which can allow some data points to stay either in between the margin area or on the incorrect side of the hyperplane.
3. What is the empirical rule?
In statistics, the empirical rule states that every piece of data in a normal distribution lies within three standard deviations of the mean. It is also known as the 68–95–99.7 rule. According to the empirical rule, the percentage of values that lie in a normal distribution follow the 68%, 95%, and 99.7% rule. In other words, 68% of values will fall within one standard deviation of the mean, 95% will fall within two standard deviations, and 99.75 will fall within three standard deviations of the mean.
4. What is the left-skewed distribution and the right-skewed distribution?
In the left-skewed distribution, the left tail is longer than the right side.
Mean < median < mode
In the right-skewed distribution, the right tail is longer. It is also known as positive-skew distribution.
Mode < median < mean
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
Company: Syngenta
Role: Data Scientist
1. What are Gaussian Mixture Model? Where they are used?
Probabilistic model for representing normally distributed subpopulations within an overall population. Mixture models in general don't require knowing which subpopulation a data point belongs to, allowing the model to learn the subpopulations automatically. Since subpopulation assignment is not known, this constitutes a form of unsupervised learning. These models have been used for feature extraction from speech data, and have also been used extensively in object tracking of multiple objects, where number of mixture components and their means predict object locations at each frame in a video sequence. change in place of 1st.
2. What are Hard-Margin and Soft-Margin SVMs?
Hard-Margin SVMs have linearly separable training data. No data points are allowed in the margin areas. This type of linear classification is known as Hard margin classification.
Soft-Margin SVMs have training data that are not linearly separable. Margin violation means choosing a hyperplane, which can allow some data points to stay either in between the margin area or on the incorrect side of the hyperplane.
3. What is the empirical rule?
In statistics, the empirical rule states that every piece of data in a normal distribution lies within three standard deviations of the mean. It is also known as the 68–95–99.7 rule. According to the empirical rule, the percentage of values that lie in a normal distribution follow the 68%, 95%, and 99.7% rule. In other words, 68% of values will fall within one standard deviation of the mean, 95% will fall within two standard deviations, and 99.75 will fall within three standard deviations of the mean.
4. What is the left-skewed distribution and the right-skewed distribution?
In the left-skewed distribution, the left tail is longer than the right side.
Mean < median < mode
In the right-skewed distribution, the right tail is longer. It is also known as positive-skew distribution.
Mode < median < mean
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
❤1
👍1
Complete Data Science Combo Course Brochure -1.pdf
10 MB
A step-by-step Roadmap towards Data Science Career.
Join our Data Science course to get complete hands on practical learning experience and become Job-Ready at best affordable price.
Download our course brochure !
Website Link👉🏻 https://www.cloudyml.com/course-combo-package/
Join our Data Science course to get complete hands on practical learning experience and become Job-Ready at best affordable price.
Download our course brochure !
Website Link👉🏻 https://www.cloudyml.com/course-combo-package/
👏1
Learn Data Visualization with Tableau from scratch and crack Data Analyst and Business Analyst Job interviews.
Actual worth price of the course is 4999/- but it is available at 2499/- today
Visit Our Webpage Now
👉 https://www.cloudyml.com/tableau/
Actual worth price of the course is 4999/- but it is available at 2499/- today
Visit Our Webpage Now
👉 https://www.cloudyml.com/tableau/
CloudyML is hiring for Data Analyst Interns
It will be paid internship. If interested, please follow the link below 👇🏻
https://forms.gle/sSLZjW6rf6X7n8sv9
It will be paid internship. If interested, please follow the link below 👇🏻
https://forms.gle/sSLZjW6rf6X7n8sv9
Today's Interview QnA
Company: Blinkit
Role: Data Scientist
Q1. You are given a train data set having 2000 columns and 2 million rows. The data set is based on a classification problem. Your manager has asked you to reduce the dimension of this data so that model computation time can be reduced. Your machine has memory constraints? How you will deal with this?
Ans:
1.Problem is because of Lower RAM - I will close all not required applications running on my machine using task manager/activity tracker - aiming to use most of memory.
2. I will try random sampling on data set. By this I mean, we can create a smaller data set, let’s say having 2000 variables and 300000 rows and do computations.
3. I will try to reduce dimensionality, I will separate numerical and categorical variables and remove correlated variables.
4. For numerical variables, I will use correlation. For categorical variables, I will use chi-square test.
5. I will also try to use PCA and pick components which can explain maximum variance in given data set
6. I guess building a Linear Model using Stochastic Gradient Descent will be helpful
7. If possible I will try to apply business understanding to estimate which all predictors can impact response variable (domain knowledge will be required). Cons can be - as this is an intuitive approach failing to identify useful predictors might result in significant loss of information
Q2. Why are Validation and Test Datasets Needed?
Ans: Data is split into three different categories while creating a model:
• Training dataset: Training dataset is used for building a model and adjusting its variables. The correctness of the model built on the training dataset cannot be relied on as the model might give incorrect outputs after being fed new inputs.
• Validation dataset: Validation dataset is used to look into a model’s response. After this, the hyperparameters on the basis of the response of the model(on the validation dataset data) are tuned.When a model’s response is evaluated by using the validation dataset, the model is indirectly trained with the validation set. This may lead to the overfitting of the model to specific data. So, this model will not be strong enough to give the desired response to real-world data.
• Test dataset: Test dataset is the subset of the actual dataset, which is not yet used to train the model. The model is unaware of this dataset. So, by using the test dataset, the response of the created model can be computed on hidden data. The model’s performance is tested on the basis of the test dataset.Note: The model is always exposed to the test dataset after tuning the hyperparameters on top of the validation dataset.
As we know, the evaluation of the model on the basis of the validation dataset would not be enough. Thus, the test dataset is used for computing the efficiency of the model.
Q3. What are Gaussian Mixture Model? Where they are used?
Ans: Probabilistic model for representing normally distributed subpopulations within an overall population. Mixture models in general don't require knowing which subpopulation a data point belongs to, allowing the model to learn the subpopulations automatically. Since subpopulation assignment is not known, this constitutes a form of unsupervised learning. These models have been used for feature extraction from speech data, and have also been used extensively in object tracking of multiple objects, where number of mixture components and their means predict object locations at each frame in a video sequence. change inplace of 1st.
Company: Blinkit
Role: Data Scientist
Q1. You are given a train data set having 2000 columns and 2 million rows. The data set is based on a classification problem. Your manager has asked you to reduce the dimension of this data so that model computation time can be reduced. Your machine has memory constraints? How you will deal with this?
Ans:
1.Problem is because of Lower RAM - I will close all not required applications running on my machine using task manager/activity tracker - aiming to use most of memory.
2. I will try random sampling on data set. By this I mean, we can create a smaller data set, let’s say having 2000 variables and 300000 rows and do computations.
3. I will try to reduce dimensionality, I will separate numerical and categorical variables and remove correlated variables.
4. For numerical variables, I will use correlation. For categorical variables, I will use chi-square test.
5. I will also try to use PCA and pick components which can explain maximum variance in given data set
6. I guess building a Linear Model using Stochastic Gradient Descent will be helpful
7. If possible I will try to apply business understanding to estimate which all predictors can impact response variable (domain knowledge will be required). Cons can be - as this is an intuitive approach failing to identify useful predictors might result in significant loss of information
Q2. Why are Validation and Test Datasets Needed?
Ans: Data is split into three different categories while creating a model:
• Training dataset: Training dataset is used for building a model and adjusting its variables. The correctness of the model built on the training dataset cannot be relied on as the model might give incorrect outputs after being fed new inputs.
• Validation dataset: Validation dataset is used to look into a model’s response. After this, the hyperparameters on the basis of the response of the model(on the validation dataset data) are tuned.When a model’s response is evaluated by using the validation dataset, the model is indirectly trained with the validation set. This may lead to the overfitting of the model to specific data. So, this model will not be strong enough to give the desired response to real-world data.
• Test dataset: Test dataset is the subset of the actual dataset, which is not yet used to train the model. The model is unaware of this dataset. So, by using the test dataset, the response of the created model can be computed on hidden data. The model’s performance is tested on the basis of the test dataset.Note: The model is always exposed to the test dataset after tuning the hyperparameters on top of the validation dataset.
As we know, the evaluation of the model on the basis of the validation dataset would not be enough. Thus, the test dataset is used for computing the efficiency of the model.
Q3. What are Gaussian Mixture Model? Where they are used?
Ans: Probabilistic model for representing normally distributed subpopulations within an overall population. Mixture models in general don't require knowing which subpopulation a data point belongs to, allowing the model to learn the subpopulations automatically. Since subpopulation assignment is not known, this constitutes a form of unsupervised learning. These models have been used for feature extraction from speech data, and have also been used extensively in object tracking of multiple objects, where number of mixture components and their means predict object locations at each frame in a video sequence. change inplace of 1st.
Q4. Say your model is suffering from lowbias and high variance !! Which algorithm you think will help you to tackle it? Why?
Ans: Low bias occurs when the model’s predicted values are near to actual values. In other words, model becomes flexible enough to mimic the training data distribution. Here not to forget a flexible model has no generalization capabilities. It means, when this model is tested on an unseen data, it gives disappointing results.
In such situations, we can use Bagging Algorithm (like random forest) to tackle high variance problem. Bagging algorithms divides a data set into subsets made with repeated randomized sampling. Then these samples are used to generate a set of models using a single learning algorithm. Later model predictions are combined using voting (classification) or averaging (regression).
Also to overcome high variance one can:
1. Use Regularization technique, where higher model coefficients get penalized, hence lowering model complexity.
2. Use top n features from variable importance chart. Maybe, with all the variable in the data set, algorithm is having difficulty in finding meaningful signal.
Ans: Low bias occurs when the model’s predicted values are near to actual values. In other words, model becomes flexible enough to mimic the training data distribution. Here not to forget a flexible model has no generalization capabilities. It means, when this model is tested on an unseen data, it gives disappointing results.
In such situations, we can use Bagging Algorithm (like random forest) to tackle high variance problem. Bagging algorithms divides a data set into subsets made with repeated randomized sampling. Then these samples are used to generate a set of models using a single learning algorithm. Later model predictions are combined using voting (classification) or averaging (regression).
Also to overcome high variance one can:
1. Use Regularization technique, where higher model coefficients get penalized, hence lowering model complexity.
2. Use top n features from variable importance chart. Maybe, with all the variable in the data set, algorithm is having difficulty in finding meaningful signal.
Machine learning interview question asked in Arraykart company. Explained solution video.
Link👉🏻 https://youtu.be/bll2mivBSqQ
Subscribe to our YouTube Channel for more such contents regularly !
Link👉🏻 https://youtu.be/bll2mivBSqQ
Subscribe to our YouTube Channel for more such contents regularly !
The demand of Cloud Computing is increasing Day by Day and Amazon Web Services (AWS) is leading the market.
CloudyML is here with a complete Hands-on course of AWS Cloud Computing in which we have covered everything which you need to know.
Visit our website and check complete syllabus and enroll for the course at best affordable price.
Course Link👉🏻https://page.cloudyml.com/AWS-course
CloudyML is here with a complete Hands-on course of AWS Cloud Computing in which we have covered everything which you need to know.
Visit our website and check complete syllabus and enroll for the course at best affordable price.
Course Link👉🏻https://page.cloudyml.com/AWS-course
Today's Interview QnA
Company: Splunk
Role: Data Scientist
1. How Are Weights Initialized in a Neural network?
Ans: There are two methods here: we can either initialize the weights to zero or assign them randomly.
Initializing all weights to 0: This makes your model similar to a linear model. All the neurons and every layer perform the same operation, giving the same output and making the deep net useless.
Initializing all weights randomly: Here, the weights are assigned randomly by initializing them very close to 0. It gives better accuracy to the model since every neuron performs different computations. This is the most commonly used method.
2. What are the variants of Gradient descent?
Ans: Stochastic Gradient Descent: We use only a single training example for calculation of gradient and
update parameters.
Batch Gradient Descent: We calculate the gradient for the whole dataset and perform the update at
each iteration.
Mini-batch Gradient Descent: It’s one of the most popular optimization algorithms. It’s a variant of
Stochastic Gradient Descent and here instead of single training example, mini-batch of samples is
used.
3. What are the feature selection methods used to select the right variables?
Ans: There are two main methods for feature selection:
Filter Methods
This involves:
• Linear discrimination analysis
• ANOVA
• Chi-Square
The best analogy for selecting features is "bad data in, bad answer out." When we're limiting or selecting the features, it's all about selecting the useful feature.
Wrapper Methods
This involves:
• Forward Selection: We test one feature at a time and keep adding them until we get a good fit
• Backward Selection: We test all the features and start removing them to see what works better
• Recursive Feature Elimination: Recursively looks through all the different features and how they pair together. Wrapper methods are very labor-intensive, and high-end computers are needed if a lot of data analysis is performed with the wrapper method.
4. What is joint sampling and separate sampling?
Ans:
· Joint sampling is done when there are equal number of events and non-events. Not appropriate for imbalanced data
· Separate sampling is done for imbalanced data. For rare event, all observations are kept when target = 1 and only few observations are kept when target = 0.
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
Company: Splunk
Role: Data Scientist
1. How Are Weights Initialized in a Neural network?
Ans: There are two methods here: we can either initialize the weights to zero or assign them randomly.
Initializing all weights to 0: This makes your model similar to a linear model. All the neurons and every layer perform the same operation, giving the same output and making the deep net useless.
Initializing all weights randomly: Here, the weights are assigned randomly by initializing them very close to 0. It gives better accuracy to the model since every neuron performs different computations. This is the most commonly used method.
2. What are the variants of Gradient descent?
Ans: Stochastic Gradient Descent: We use only a single training example for calculation of gradient and
update parameters.
Batch Gradient Descent: We calculate the gradient for the whole dataset and perform the update at
each iteration.
Mini-batch Gradient Descent: It’s one of the most popular optimization algorithms. It’s a variant of
Stochastic Gradient Descent and here instead of single training example, mini-batch of samples is
used.
3. What are the feature selection methods used to select the right variables?
Ans: There are two main methods for feature selection:
Filter Methods
This involves:
• Linear discrimination analysis
• ANOVA
• Chi-Square
The best analogy for selecting features is "bad data in, bad answer out." When we're limiting or selecting the features, it's all about selecting the useful feature.
Wrapper Methods
This involves:
• Forward Selection: We test one feature at a time and keep adding them until we get a good fit
• Backward Selection: We test all the features and start removing them to see what works better
• Recursive Feature Elimination: Recursively looks through all the different features and how they pair together. Wrapper methods are very labor-intensive, and high-end computers are needed if a lot of data analysis is performed with the wrapper method.
4. What is joint sampling and separate sampling?
Ans:
· Joint sampling is done when there are equal number of events and non-events. Not appropriate for imbalanced data
· Separate sampling is done for imbalanced data. For rare event, all observations are kept when target = 1 and only few observations are kept when target = 0.
.................
🔴If you want to learn Data Science from scratch and crack interviews, join our Assignment based Beginner to Advanced Data Science Combo course, trusted by 5000+ Learners.
It Covers Python, ML, DL, SQL, Statistics and Maths, AWS Sagemaker, Model Deployment, 25+ Capstone Projects for ML and DL, Interview QnAs PDF Collections and comes with 1-1 Live Doubt Clearance Support on skype chat.
Visit Course Now 👉🏻 https://www.cloudyml.com/course-combo-package/
If you wanna make your career in Data Science, join our Data Science Combo Course Today.
Save your lakhs of rupees and learn at best affordable price.
We have more than 5000 happy learners with a lot of success stories that you can check on our website.
This course covers everything from basics to advanced and anyone from any background can join and switch their career into Data Science
Website Link👉🏻 https://www.cloudyml.com/course-combo-package/
Course Brochure👉🏻 https://ai.cloudyml.com/brochure
🔴Enroll Today at discounted price
Save your lakhs of rupees and learn at best affordable price.
We have more than 5000 happy learners with a lot of success stories that you can check on our website.
This course covers everything from basics to advanced and anyone from any background can join and switch their career into Data Science
Website Link👉🏻 https://www.cloudyml.com/course-combo-package/
Course Brochure👉🏻 https://ai.cloudyml.com/brochure
🔴Enroll Today at discounted price