MAI272 – Advanced Machine Learning

Program_1: Simple Linear Regression:

Each Sample in the dataset describes a house in Boston town. The data was drawn from the Boston
Standard Metropolitan Statistical Area (SMSA) in 1970. The attributes are defined as follows

● Crime_Rate per capita crime rate by town
● Zone_proportion of residential land zoned for lots over 25,000 sq.ft.
● Non_Retail_Acres proportion of non-retail business acres per town
● CHAS Charles River dummy variable (= 1 if tract bounds river; 0 otherwise)
● NOX nitric oxides concentration (parts per 10 million)
● Avg_rooms average number of rooms per dwelling
● AGE proportion of owner-occupied units built prior to 1940
● Distance weighted distances to five Boston employment centres
● RAD index of accessibility to radial highways
● Property_TaxRate full-value property-tax rate per 10 000 USD
● PTRATIO pupil-teacher ratio by town
● B 1000 (Bk - 0.63)^2 where Bk is the proportion of black people by town
● LSTAT % lower status of the population
● MValue Median value of owner-occupied homes in $1000's

Use the dataset and perform the following activities:

1. Consider the column, ‘Avg_room’, as predictor, and ‘MValue’ as the target variable
2. Visualize the association between the predictor and the target using scatter plot.
3. Split the data into train and test datasets, in the ratio of 70:30.
4. Build a Linear Regression model using training dataset, to predict the target variable.
5. Observe the coefficient and intercept values for the model.
6. Evaluate the model using mean squared error values and R-squared values on the training and the testing datasets.
7. Provide suitable inference on the model created.

Note: Dataset can be pre-processed if required

Evaluation Rubrics

Content Score

Explanation / Equation / Worked-out examples for demonstrating the flow of the algorithm

3 Marks

Use of appropriate functions / parameters 5 Marks Concept Clarity (Algo & Dataset) 2 Marks Total 10 Marks

Department of Computer Science

MAI272 – Advanced Machine Learning

Program_2: Multiple Linear Regression

Each Sample in the dataset describes a house in Boston town. The data was drawn from the Boston
Standard Metropolitan Statistical Area (SMSA) in 1970. The attributes are defined as follows

● Crime_Rate per capita crime rate by town
● Zone_proportion of residential land zoned for lots over 25,000 sq.ft.
● Non_Retail_Acres proportion of non-retail business acres per town
● CHAS Charles River dummy variable (= 1 if tract bounds river; 0 otherwise)
● NOX nitric oxides concentration (parts per 10 million)
● Avg_rooms average number of rooms per dwelling
● AGE proportion of owner-occupied units built prior to 1940
● Distance weighted distances to five Boston employment centres
● RAD index of accessibility to radial highways
● Property_TaxRate full-value property-tax rate per 10 000 USD
● PTRATIO pupil-teacher ratio by town
● B 1000 (Bk - 0.63)^2 where Bk is the proportion of black people by town
● LSTAT % lower status of the population
● MValue Median value of owner-occupied homes in $1000's

Use the dataset and perform the following activities:

1.Consider the attributes Av_room, distance, Property_TaxRate, Non_Retail_Acres as predictors and
MValue as the target variable.
2.Calculate the Variance Inflation Factor for each of the selected predictors. Based on the VIF factor,
finalize the list of predictors
3.For every predictor identified, visualize its association with the target column using scatter plot.
4.Split the data into train and test datasets, in the ratio of 67:
5.Build a multiple linear regression model to predict the target variable using the selected predictors
6.Observe the coefficients and intercept values for the model
7.Evaluate the model using mean squared error values, R-squared Values and adjusted R-squared
values on the train and test dataset
8.Provide suitable inference on the model created.

Note: Dataset can be pre-processed if required

Evaluation Rubrics
Content Score

Explanation / Equation / Worked-out examples for demonstrating the flow of the algorithm

3 Marks

Use of appropriate functions / parameters 5 Marks Concept Clarity (Algo & Dataset) 2 Marks

MAI272 – Advanced Machine Learning

Program_3a: Logistic Regression:

Perform the following using Program 3a_dataset

1. Consider duration, age, and campaign columns as predictors and the column Y as target variable. Y says whether the client will subscribe a term deposit or not.
2. Use 70% of the data as training data set and 30% of data as testing data set.
3. Build a logistic regression model.
4. Determine the classification accuracy rate of the model on both train and test set.
5. Display Classification Report and Confusion Matrix
6. Provide suitable inference on the model created.

Program_3b: Logistic Regression:

Perform the following using Program3b_dataset

1. Consider 'variance', 'skewness', 'curtosis', 'entropy' columns as predictors and the column 'class' as the target variable
2. Build a logistic regression model.
3. Calculate the classification accuracy, precision and recall for the model
4. Display Classification Report and Confusion Matrix
5. Provide suitable inference on the model created.

Note: Dataset can be pre-processed if required

Evaluation Rubrics
Content Score

Explanation / Equation / Worked-out examples for demonstrating the flow of the algorithm

3 Marks

Use of appropriate functions / parameters 5 Marks Concept Clarity (Algo & Dataset) 2 Marks

MAI272 – Advanced Machine Learning

Program_4a: Polynomial Regression:

Use the program_4_dataset attached and perform the mentioned activities.

Dataset description:

The dataset contains outdoor temperature and pressure data between the duration as given in the
timestamp. The dataset is presented in OS_time, temperature, pressure humidity and actual date and
time format. Interesting patterns between temperatures, pressure can be found for understanding the
humidity or rainfall during the given period. This can also be used to understand environmental
changes over time.

1. Consider Temperature as predictor and the target as pressure
2. Build a model using polynomial regression with at least 6 different degrees with 3 different train-test split as 66-44, 70-30, 80-20.
3. Calculate R2 square. Mean Squared Error, Mean Absolute Error and validate your model
4. Provide suitable inferences on the model created.

Note: Dataset can be pre-processed if required

Evaluation Rubrics
Content Score

Explanation / Equation / Worked-out examples for demonstrating the flow of the algorithm

3 Marks

Use of appropriate functions / parameters 5 Marks Concept Clarity (Algo & Dataset) 2 Marks

MAI272 – Advanced Machine Learning

Program_5: Lasso and Ridge Regression:

Dataset description:

All attributes except Category and Sex are numerical. The laboratory data are the attributes 5-14.

1. X (Patient ID/No.)
2. Category (diagnosis) (values: '0=Blood Donor', '0s=suspect Blood Donor', '1=Hepatitis', '2=Fibrosis', '3=Cirrhosis')
3. Age (in years)
4. Sex (f,m)
5. ALB
6. ALP
7. ALT
8. AST
9. BIL
10. CHE
11. CHOL
12. CREA
13. GGT
14. PROT

Perform the following using Program_5_dataset

1. Build a simple linear Regression considering ‘CREA’ as predictor variable and ‘PROT’ as Target variable (for all models)
2. Calculate the VIF for all the attributes except X (Patient ID/No.), 2) Category 3) Age, and 4) Sex. Based on the VIF factor, finalize the list of predictors
3. Split the data into train and test datasets in the ratio of 80:20 (for all experiments)
4. Build a multiple linear regression model to predict the target variable ‘PROT’ using the selected predictors
5. Build a polynomial regression model using the attribute that has highest VIF for 3 different degrees
6. Build a Ridge regression model using the attribute that has highest VIF for 13 different values of alpha ranging from 0.1 to 150
7. Build a LASSO regression model using the attribute that has highest VIF for different values of alpha ranging from 0.1 to 100
8. Use N-fold cross validation for training the model created.
9. Evaluate the models using mean squared error values, R-squared Values and adjusted R-squared values (wherever applicable) on the train and test dataset
10. Print intercept and coefficient values wherever appicable
11. Provide a comparative analysis graph of models built and give conceptual inferences on all the models built

Note: Dataset can be pre-processed if required Evaluation Rubrics

Content Score
Explanation / Equation / Worked-out
examples for demonstrating the flow of
the algorithm

3 Marks

Use of appropriate functions / parameters 5 Marks
Concept Clarity (Algo & Dataset) 2 Marks