Model Lifecycle Class 12 Questions and Answers

These Class 12 AI Important Questions Chapter 2 Model Lifecycle Class 12 Important Questions and Answers NCERT Solutions Pdf help in building a strong foundation in artificial intelligence.

Model Lifecycle Class 12 Important Questions

Class 12 AI Model Lifecycle Important Questions

Important Questions of Model Lifecycle Class 12 – Class 12 Model Lifecycle Important Questions

Model Lifecycle Class 12 Very Short Answer Type Questions

Question 1.
What do you mean by data set?
Answer:
A data set is a collection of data in which data is arranged in a particular order. It may contain a series of data or a single data.

Question 2.
Why data set is needed?
Answer: As the machine learning model project is totally based on data. Without which we cannot train or built AI model.

Question 3.
Do you think the ML project can work properly if the data sets is not well prepared or preprocessed.
Answer:
No

Question 4.
What Are the three stages of building a machine learning model.
Answer:
The three stages of building a model are :

• Model building through a suitable algorithm and training of the model
• Model accuracy testing through test data
• Applying the model after testing and to use it in real time project

Question 5.
What do you understand by training set and test set in a machine learning model?
Answer:

• Training set : The data set given to the model to analyze and learn it is the 70 % of the total training data set. This is usually the leveled data.
• Testing set: This data set is used to test the accuracy of the model. The remaining data set after training i.e. 20-30 % is taken as testing data set. Usually testing data set is unlabeled data.

Question 6.
When evaluation stage of the model is performed?
Answer:
After a model has been created and trained it needs to be properly tested to calculate the models efficiency and performance. A separate validation data set is used for model evaluation, that generalizes how well the model is behaving, by avoiding bias and over fitting.

Question 7.
Which stage of the model life cycle consumes maximum time?
Answer:
The data collection and data exploration should be done in a carefully.

Question 8.
What are the stages in the phase one of stage of model life cycle.
Answer:
Problem scoping, data acquisition and data exploration.

Question 9.
In which stage modeling and evaluation of the model is performed?
Answer:
In the second stage or phase two modeling or evaluation is performed.

Question 10.
What is scoping of a problem?
Answer:
It can be defined as the deeper understanding of the problem and it gets clear when we attempt it to solve it.

Question 11.
What is 4 WS problem canvas?
Answer:
4 WS problem canvas are the problem statement that can be formulated as follows:

Question 12.
What are the issues that one may encounter in the collected data?
Answer:
Some of the issues that may encounter with the collected data are:

• Missing values
• Duplicate data
• Invalid data
• Noise

Question 13.
What is outliers in machine learning?
Answer:
Outliers are those data points that are significantly different from the rest of the dataset. They are often irrelevant observations that skew the data distribution, and arise due to inconsistent data entry, or erroneous observations.

Question 14.
What do you think the best common way to handle missing data?
Answer:
The best common way is either remove the whole record of data or insert a value that is quite close to the missing data.

Question 15.
What are the types of data used in data sets?
Answer:

• Numerical data
• Categorical data
• Ordinal data

Model Lifecycle Class 12 Short Answer Type Questions

Question 1.
What are the considerations to be taken during the evaluation stage?
Answer:
These are the few points to be taken into considerations:

• The volume of test data should not be huge as it may provide data complexity.
• Some times the system may deal with sensitive data so regulatory compliance and security testing are essential.
• It is also necessary for system integration testing, if the model the require data from other systems.
• The testing data must be developed by the team who are involved in the validation of the ML models.

Question 2.
“Once the relevant projects have been selected and properly scoped, the next step of the machine learning lifecycle is the Design or Build phase.” Briefly explain this phase.
Answer:
Data gathering, investigation, preparation, cleaning, feature engineering, testing, and running a number of models in an attempt to anticipate behaviors or find insights in the data make up the majority of the steps involved in constructing an AI or machine learning model.

Model Lifecycle Class 12 Long Answer Type Questions

Question 1.
What do you understand by the term over fitting, under fitting and model fit in terms of model testing.?
Answer:
In terms of modeling over fitting correspondence to the predictions that the model predicts on testing data and it is due to the inaccurate prediction of the model. Either due to the to much detailing or due to the noise.

The reason is the model is too complex.
Under fitting : A machine algorithm is said to be under fitting when jt performs well on the training data but no on the testing data. This is due to the fewer data to build the model or due to non-linear data. That is the reason the model makes wrong predictions. Another reason is model is to simple and size of training data set not enough.

Model fit: When a model is best fitted it produces more accurate outcomes and it is essential because if the model does not fit your data correctly the outcomes will not be accurate enough to be useful for practical decision making.

Question 2.
What are the challenges in the designing stage of model.
Answer:
One of the major challenges is to is in the development process. Some of the major challenges are :

• Increasing complexity
• Adaption to the new approach
• Automated model verification
• Model based testing

Question 3.
What are the steps involved in data cleaning?
Answer:
The various steps use for data cleaning are :
Removal of duplicate observations: There are many chances of duplicate observations in the collected data from multiple places or from online sources, so it is necessary to remove those duplicate, otherwise it may effect the model building.

Removal of irrelevant observations: Sometimes the data set collected may have lots of extra irrelevant data that may does not add value to the problems and it is necessary to remove them as the model has to learn from good relevant data.

Removal of unwanted outliers: Unwanted outlier removal is necessary so that it won’t effect the models performance.

Handling missing data: In many cases there are chances of missing data with the data sets for machine validations. Sometimes machines don’t work well with missing data. So it should be handled at the cleaning stage.

Model Lifecycle Class 12 Notes

Machine learning life cycle: It is the step by step cyclical process that’AI and machine learning follows. It mainly comprised three main stages.

• Stage 1 : problem scoping
• Stage2:Design or building phase
• Stage3: Deployment and maintenance in production

Requirement analysis: This is the most important part of ne AI project, Due to two reasons. First as it involves the planning and motivational aspects of project.

Second stage of requirement analysis: Second stage is the data acquisition. As it characterizes the project stage : garbage in, garbage out i.e. if the collected data is not good an effective AI model can be found.

Data acquisition: In order to make a effective model it is very much necessary to collect data from credible sources. Most likely real world data can be strange and deceptive. As chances of Human errors is more likely i.e. miss typing of 400 as 40 or 40.40 or smelling mistake for that reason authentic source of data is required and can be collected from the following sources.

• Government websites
• Devices such as cameras and sensors
• Purchases, transactions, registrations, and
• Other public surveys and records

Widely used sources of machine learning data source: List of various data sets given below:

• Kaggle datasets
• UCI machine learning repository
• Datasets via AWS
• Google’s dataset search engine
• Microsoft datasets
• Awesome public dataset collection
• Government datasets
• Computer vision datasets
• Scikit-learn dataset

Data exploration: Once the data is collected then the data is scrutinized for appropriate needs and interest. Then relationship and patterns within the data explored. In order to visualize the data quickly using different visualization tools, to get a sense of the trend and relationship within the data.

Various visualization or graphical representations: Graphical representations, flowcharts, hierarchies and process and process descriptors etc can be seen here.

• Bubble map
• Calendar
• Donut chart
• Dot map
• Dot matrix chart
• Bubble chart

Stage 2 design and testing the AI model: Once the required project has been selected and properly scope the next step of the model life cycle is the design/ or building phase. It may take days to month depending on the nature of the project. It comprises of the following steps.

• Data acquisition
• Exploration
• Preparation
• Cleaning
• Feature engineering
• Modelling
• Evaluation

Deployment and maintenance: This is the final stage of model life cycle. Here the model is monitored to ensure that it ful fill all the requirements of the business. As there are chances of several detrimental effect that may occur overtime, model deterioration most common one. In this stage second step is feedback. Collected from different sources after using the model.

The post Model Lifecycle Class 12 Questions and Answers appeared first on Learn CBSE.