Machine Learning Algorithms and Applications M6106

Assignment:

This assignment requires you to run multiple experiments with the following instructions:

Experiment 0 :

Define the business use case and users for your models.

iuww520iuww520iuww520iuww520iuww520iuww520iuww520iuww520

Perform exploratory data analysis on the dataset.

Apply data transformation, data cleansing and feature engineering required for training ML models Pick any number of numerical and/or categorical features of your choice from the dataset Assess the performance of a baseline model using appropriate metrics

Experiment 1 :

Use the prepared data sets from experiment 0

Use the list of chosen features from experiment 0

Train several Multivariate Linear Regression models with different values for the hyperperparameter: fit_intercept Analyse their results and experiment’s outcomes Make recommendations for next experiment (for example: more required data preparation, recommendation on hyperperparameter, …)

Experiment 2 :

Use the prepared data sets from experiment 0

Use the list of chosen features from experiment 0

Train several Ridge Linear Regression models with different values for the hyperperparameter: alpha Analyse their results and experiment’s outcomes

Make recommendations for next experiments.

Experiment 3 :

Use the prepared data sets from experiment 0

Use the list of chosen features from experiment 0

Train several Lasso Linear Regression models with different values for the hyperperparameter: alpha Analyse their results and experiment’s outcomes

Make recommendations for next experiments.

For each experiment, you will need to fill out a notebook containing your Python code but also relevant information such as hypothesis, chosen approach, rationale behind decisions made, results achieved and issues faced, insights gained and recommendations. Each section that requires your input has been marked with the tag <fill_this>.

Some code has already been implemented in the notebooks. You should not modify them and should use the defined variables.

At the end of the experimentation phase (i.e. after completing all experiments), you will write a final report presenting the business use case that will benefit from your machine learning model predictions, the dataset used for training, the approaches you took during your experimentation phase, best results achieved, analysis on correct and incorrect predictions, business impacts and ethical consideration of your model and final recommendation. The report should not exceed 3000 words. You will use the following project report template: