Machine Learning

What is Machine Learning?

Machine Learning is a subfield of Artificial Intelligence (AI). It enables machines to learn from past experiences or historical data without being programmed explicitly. It was first introduced by Arthur Samuel in 1959 as a field of study that allowed learnings for computers without specifically programming them.

Tom Mitchell in 1998 gave an idealistic definition to this statement as a Machine Learning Problem - "A computer program is said to learn from experience 'E' with respect to some task 'T' and some performance measure 'P', if its performance on 'T', as measured by 'P', improves with experience 'E' ".

Algorithms in Machine Learning:

Algorithms in ML mainly include: -

  • Classical Learning

  1. Supervised Learning

  2. Unsupervised Learning

  • Ensemble Methods

  1. Bagging

  2. Boosting

  • Reinforced Learning

  • Neural Networks and Deep Learning

  • Recommender Systems

Most of the machine learning products are fabricated with already existing ML algorithms with some minor tweaks and changes.

Of a wide variety of machine learning algorithms, three main categories groups are Supervised, Unsupervised, and Reinforcement Learning.

In order to use the algorithms correctly, it is important to know the life cycle of Machine learning.

Machine Learning Life Cycle:

There are majorly seven definitive steps involved in the entire life cycle of machine learning:

  1. Data Gathering (data sources, data collection, and integration)

  2. Data Preparation (data exploration, data pre-processing)

  3. Data Wrangling (data cleaning and data transforming)

  4. Data Analysis (building models and reviewing results)

  5. Training models

  6. Testing models

  7. Finally, Deploying the model

Case Studies:

Employee Attrition [Corporate Personnel] Project Link

Business Objective: Aim was to know what makes an employee quit or leave

Solution: Understand factors involved in influencing employees using a Regression model

Key Achievement: Developed a model using Logistic Regression to highlight significant attrition attributes

Future-500 [Profit Makers] Project Link

Business Objective: Aim was to analyze how does the company like Future-500 marks its growth

Solution: Understand key variables on scale of revenue and expenses to predict success/failure

Key Achievement: Designed a Regression model using Linear Regression to predict business profitability

Mobile Phone Usage [Trait Classification] Trait Classification : rMarkdown

Business Objective: Aim was to understand the data points upon similar features but different Prices

Solution: Performed K-means clustering to find and compare the most affected values

Key Achievement: Performed EDA using Cluster Analysis to customer behavior and market position

Grocery Cart [Sales Markers] Sales Markers : html-notebook

Business Objective: Aim was to identify which products have a higher market and why

Solution: Interpret sales correlations on transactional data for the groceries

Key Achievement: Performed Basket Analysis using Applied Association Rule Mining to identify valuable variables for sale promotions

Apple Phone Review Comments [Intonation Classification] Sentiment Analysis : rMarkdown

Business Objective: Aim was to understand the positive and negative comments given in reviews

Solution: Performed Sentimental Analysis in a supervised environment to gain insight into the customer experience of the product

Key Achievement: Performed regression using SVM to understand customer stances and feedback