You are on page 1of 3

Mini Project guidelines – MBA-DSDA DSDA

Course: Machine Learning, Faculty: Dr. Angelina Gokhale


Instructions:
1. Each group is required to fill up that questionnaire for the mini project [you have already done that]
2. Follow the process of building an end-to-end machine learning project on the use cases assigned
(Refer the attached PPT for a quick reference on the steps, omit the deployment phase)
3. Explore the dataset given and make your analysis on the models (1 for each member) that you may
select on the dataset (feel free to change your target variable if required)
4. Now, describe your problem statement in ChatGPT to receive suggestions on the different ML
models that are applicable to your use case. (Attach a screenshot of the answer obtained).
5. Build your model using any 3 of those suggestions obtained from ChatGPT and justify the same.
6. If your approach wasn’t in the list provided by ChatGPT, then feel free to use the model you think
works best on your problem statement
7. Your approach to the problem statement will be given a higher priority in grading the same.
8. The following instructions on the presentation preparation will guide you through the important
details to be focused upon while working on this project.
9. For submission, submit your Jupyter Notebook (either Python or R) (email the same to
angelina.gokhale@associates.scit.edu)
 prepare one Jupyter notebook combining every members’ contribution - rename that notebook
with your group number and email it to me (deadline will be mentioned)
 [Naming convention for the File name: <Group No.>_ProjectTitle_MiniProject.ipynb, e.g.
1_CreditCardFraudDetection_MiniProject.ipynb]
 also save that notebook (.ipynb) as a PDF file and submit the same.
 [Naming convention for the File name: <Group No.>_ProjectTitle_MiniProject.pdf, e.g.
1_CreditCardFraudDetection_MiniProject.pdf]
 both the .ipynb and .pdf files have to be emailed to me along with your presentation
(presentation can be submitted after your project viva)

Presentation:
1. Title Slide (1 Slide)
2. Dataset Description (1-2 Slide) [includes the domain details, competition details (if any) that this
dataset has been used for, summary of results of previous execution on this dataset] Mention the
train-test details—exploration of the data – correlation plots/ histograms or any other relevant
exploration.
3. Use cases modeled for this dataset (2-3 Slides)
4. Details of the models selected for this problem statement along with the necessary justification (1-2
slides)
5. Details of fine-tuning the model and the regularization techniques if implemented (2 slides)
1. try out GridSearchCV
2. search and try out regularized forms of the model (e.g. Ridge and Lasso regression for regularized
linear regression models)
6. Results (1-2 slides)
3. Prepare a comparative chart of the models tested and any details of fine-tuning the model
4. Here, mention the use case, model used, details of the hyperparameters, no. of iterations (for
training) and the evaluation metrics

1
Mini Project guidelines – MBA-DSDA DSDA
Course: Machine Learning, Faculty: Dr. Angelina Gokhale
7. Conclusion based on your findings (1-2 slides)
5. which model performed well and which technique is suitable to address this dataset/ problem area
6. comments on the bias and variance of your model
7. Learning from the project (1 slide)
8. each member is supposed to mention individual learning
9. Final thank you slide (1 slide)

2
Mini Project guidelines – MBA-DSDA DSDA
Course: Machine Learning, Faculty: Dr. Angelina Gokhale

You shall be graded as follows:


1. Organization of Kaggle workbook / Jupyter notebook (10marks)
2. Explanation of use case (5marks)
3. Q&A (5marks)
4. Title & thank you slide (1mark)
5. Dataset description (5marks)
6. Model selection and justification (15marks)
7. Methods to improve the model (fine-tuning and regularization) (5marks)
8. Results (10marks)
9. Conclusion (3 marks)
10. Learning (1mark)

The total project score will be scaled down to 35 marks.

You might also like