Machine Learning Fundamentals Lab

This repository contains Jupyter notebooks used for training during the hands-on session.

Contents

• 01-0_NumPy Quickstart.ipynb: Create matrics and apply mathematical operations on it.
• 01-1_Linear Algebra with NumPy.ipynb Discussion of basic matrix operations and Gaussian elimination.
• 01-2_Statistics and Probability.ipynb Fundamentals of probability, probability distributions, Bayes' theorem and the Central Limit Theorem.
• 02-0_Data Exploration and Data Transformation: Explore a dataset and perform data visualization and data transformation.
• 03-0 Feature Scaling: Perform feature scaling.
• 04-0_Linear Regression 1: Predict the house price by using Linear Regression.
• 04-1_Linear Regression 2: Predict the stock price by using Linear Regression.
• 05-0_Logistic Regression : Perform Binary Classification by using Logistic Regression.
• 06-0_Support Vector Machines (SVM) : Perform Multiclass Classification by using Support Vector Machines (SVM).
• 07-0_Decision Tree and Random Forest : Perform Multiclass Classification by using Decision Tree and Random Forest.
• 08-0_K Nearest Neighbour (KNN) : Perform Multiclass Classification by using K-Nearest Neighbours.
• 09-0_K Means Clustering 1: Perform Unsupervised Clustering by using K-Means Clustering.
• 09-1_K Means Clustering 2: Use K Means Clustering to reduce colour space of an image.
• 10-0_Principal Component Analysis (PCA): Perform Dimension Reduction and Data visualization by using Principal Component Analysis (PCA).

Dependencies

Preferred kernel and IDE

Python 3.7.3 or newer and Jupyter Notebook

If you already have Anaconda, you can create a new virtual environment with Python 3.7.3 by and name your environment name

conda create --name <REPLACE-NAME> -y

Go into your newly created python3.7.3 environment by

conda activate <REPLACE-NAME>

Deactivate the newly created environment using

conda deactivate

Installation on Windows 10

If you do not have Python on your machine, download miniconda from here.

After installing miniconda, install the required python modules required for this hands-on session.

conda install --file requirements.txt -y

If you already have python on your machine, try installing the python modules and proceed only if all modules are installed successfully.

python -m pip install --r requirements.txt

All modules will be installed successfully except OpenCV installation will fail. Install manually using

python -m pip install opencv-python==3.4.2.16

Getting started

In Anaconda Prompt (Windows systems) or your system terminal (non-Windows systems), launch Jupyter Notebook by

jupyter notebook

and go to the directory where you have downloaded this repository.

OR directly open Jupyter notebook server at the PATH with the downloaded repository by

jupyter notebook <YOUR-PATH-DOWNLOADED-REPOSITORY>