Coding Club Coding Week 24
IIT Guwahati
Machine Learning
Task
Deadline on
20.05.24
Machine Learning Task
Greetings
Hello there!
Hope you had a fun time going through the resources shared. Don’t worry if you were unable to
complete them, you can always refer to them whenever you require while doing the tasks. We
have planned 2 tasks to judge your understanding and creativity :)
Task 1 will be based on your creativity to see if you can look at data and think of interesting
questions, Task 2 will be based on implementation of some ML algorithms on a dataset.
Good luck, and hope you have a good learning experince.
Task 0: Setup Environment
This is related to setting up your VS-Code environment to enable you to complete the tasks in a
Jupyter Notebook, you can skip this if you have already jerry-rigged your Jupyter Notebook
development environment
Download VS-Code from here
Install Python from here
Install pip (Python’s Package Manager
Windows: Find instructions here
Linux: Run in terminal: sudo apt install python3-pi
Mac: Run in terminal:
curl [Link] -o get-pip.p
python3 get-pip.p
Run in terminal:
pip install ipykerne
pip install noteboo
Install the following VS-Code Extensions
Pytho
Jupyte
Jupyter Key Ma
Jupyter Notebook Renderers
Task 1: Dataset Analysis
Objective
The objective of this task is to perform data cleaning and exploratory data analysis (EDA) on the
campus dataset. The dataset contains various columns such as year of study, department,
relationship status, and more. Your goal is to clean the data and then derive meaningful insights
through EDA.
Machine Learning Task
Instructions
Step 0: For this task you are required to download this csv file first.
Step 1: Data Loading and Initial Cleanin
Load the [Link] file into a pandas DataFrame
Drop the ‘Timestamp’ column as it will not be used in the analysis
Rename the columns to shorter names for easier reference.
Step 2: Data Cleanin
Refer to this notebook to clean this dataset
Explore additional ways to clean the data (for example, clean redundant values in the
favorite sport column).
Step 3: Exploratory Data Analysis (EDA
After cleaning the dataset, proceed to EDA to interpret the relationships between various
columns or gain insights into individual columns
To guide you, some example plots and relationships on the dataset are shown in this
notebook
If you need further reference for EDA, you can watch this video
Now, your task is to think of and answer 5 more questions by drawing appropriate
plots and also write inferences drawn from these plots.
Judging Criteria: Creativity in EDA, some extra points for those who do Data Cleaning,
Originality (there will be a strict plagiarism check).
Submission: Name of File <roll_number>_1.ipynb. Submit the Google Drive link of your
prepared notebook (Properly comment your notebook with markdown blocks, for each of the
steps you do). The submitted notebook should include all steps starting from Step 1, then data
cleaning as done in the shared notebook and any further data cleaning you do. After that include
the EDA, questions, plots, analysis and inference.
Task 2: Regression
In this task you will explore regression by trying to predict the number of 4’s, on the shared IPL
Dataset.
The dataset can be downloaded from here
Read the csv datafile as a Pandas DataFram
Clean up the data, some possible things you can do (try to explore more) are, seperate
player names from team name, check for NULL values and handle them if present, etc
Plot graphs and explore various correlations in the dataset, for example try to explore how
the number of 4’s depend on strike rate or some other factor.
Machine Learning Task
The target column is the number of 4’s the rest of the columns are input features, define two
new dataframes X and Y accordingly, you can use some form of feature selection here to
select good features for your X
Split the dataset using SKLearn train_test_split, with random_state = 42 and test size = 0.3
(for reproducibility
Train a Linear Regression model on your train data, and make predictions on the test data,
you need to find mean squared error for both train data predictions and test data predictions
Once you have trained Linear Regression, you can explore other models to try to decrease
your MSE on the test dataset, some other models you can try are Polynomial Regression,
Decision Tree Regressor, Random Forest Regression, etc. (As of now, you just need to
implement various models using SKLearn, you can explore the theoretical foundations later)
(If you want some help you can checkout this Kaggle Notebook to learn how to implement
Linear Regression with SK Learn
Report your Mean Squared Error on the test dataset (generated using train test split) for any
model you try.
Judging Criterion: Feature Selection, Best MSE Score, Quality of approaches tried, Clarity and
Code Structure, EDA
Submission: Name of File <roll_number>_2.ipynb, upload the notebook to Google Drive
(allow view access and share link in form). Write properly commented code separated by
markdown blocks.
Deadline - 20th May
Submission Link
Join the WhatsApp Group
Have any doubts ?
Contact us
Adarsh +91 96509 46587
Coding Club
Mahek +91 86198 90965 IIT Guwahati