Search
  • ankitrathi

The Machine Learning Curve

Source: http://stamfordresearch.com/


In the current era of open internet & MOOCs, its pretty easy to be perplexed by plethora of options available & its a challenge to decide from where to get information and grasp it in the most optimized way. Machine Learning beginners can easily relate to it, so many books, courses, lectures, presentations available on the internet but very little information on where to start. In 2012, when I started with my ML journey, not many options were available but things have changed a lot in last 4 years.

So, here is an attempt to consolidate the information available on the internet in a structured way (with links to useful sources) for a ML starter to kick-off his journey based on my experience as a learner.

Note: As ML is a vast area of application & research these days, not each and every aspect is covered here, its just an approach of learning for ML starter and enable him to explore further in a structured & optimized way.

1. What & Why of Machine Learning

First & foremost, you should be starting with exploring what is ML & why it is so useful these days? What are the types of ML and which type serves what kind of problem?

-Machine Learning

-ML Types

— Supervised

— Unsupervised

— Reinforced

Links

Machine Learning with R — Brett Lantz

Machine Learning on Coursera -Andrew Ng

Machine Learning Mastery -Jason Brownlee

2. Basics of R & Python

After getting the introduction to ML, you should have a look at the tools (R/Python etc) for the basics of programming & data manipulation so that you can have a look at sample solutions or tutorials to get a view how ML problems are solved using these tools.

-R

— Basic Syntax

— Condition & Loops

— Data Manipulation

Links

Machine Learning with R — Brett Lantz

Learn Data Science in R — Analytics Vidhya

-Python

— BasicSyntax

— Condition & Loops

— Classes & Methods

— Data Manipulation

Links

Python Machine Learning — Sebastian Raschka

Learn Data Science in Python — Analytics Vidhya

3. Kaggle Tutorials

Once you are done with the tools part, you can have a look at some basic tutorials at Kaggle, the problem mentioned below covers different domains of ML (mainly classification & regression). Its important to get your hands dirty quickly with practical problems so that you can stay motivated & get the context when you dive deep into the theory.

Titanic Problem

House Pricing Problem

Digit Recognizer

4. Maths/Stats Basics

Before starting with what, why & how of different ML algorithms, I would advice you to get started with the basics of Statistics, Probability & Linear Algebra, which will prepare you to grasp the working on algorithms without any discomfort.

Harvard Statistics 101

Probability Theory

Linear Algebra for Machine Learning

5. Machine Learning Algorithms

Now, in my opinion, is the right time to understand & dive deep into the ML algorithms with a view on what algorithm solves what kind of practical problem?

-Supervised

— Classification

— Regression

— Common

-Unsupervised

— Association

— Clustering

— Anomaly Detection

-Reinforced

Links

Essentials of Machine Learning Algorithms -Analytics Vidhya

List of machine learning concepts -Wikipedia

6. Machine Learning Process/Framework

Once you are familiar with all the topics above, I believe you can now have a look at the complete ML process/framework. Here you will learn, how to collect data, how to understand, explore & prepare data for ML algorithms to be consumed, how to evaluate the performance & how to improve the performance of the model.

-Data Collection Methods

-Exploratory Data Analysis -Analytics Vidhya

-Applying Models/Algorithms

— Linear Regression

— Logistic Regression

— Decision Tree

— Support Vector Machines (SVM)

— Naive Bayes

— KNN

— K-Means

— Random Forest

— Dimensionality Reduction Algorithms

— Gradient Boost & Adaboost

-Evaluation with Cross-Validation

Improve Your Model Performance using Cross Validation -Analytics Vidhya

-Improving Model Performance

— Feature Engineering

— Ensembling

— -Bagging

— -Boosting

— -Stacking

Links

How to build Ensemble Models in machine learning? -Analytics Vidhya

Kaggle Ensembling Guide -MLWave

7. Machine Learning Practice

So now, you know the theory, you know the tools, you know the process/framework, its the right time to get your hands dirty with practical problems and now its the matter of practice, practice & practice to excel in ML space.

-Kaggle Competitions

-Hackathons

-On Job

I hope you liked the article & it proves useful to your journey, it may give some insights to you so that you can develop your own learning path, happy learning.

Thank you for reading my post. I regularly write about Data & Technology on LinkedIn & Medium. If you would like to read my future posts then simply ‘Connect’ or ‘Follow’. Also feel free to connect on Slideshare.

Originally published at https://www.linkedin.com on March 21, 2017.

#DataScience #MachineLearning

1 view

Call

T: +91 9891XXX969  

Follow me

  • Facebook Clean
  • Twitter Clean
  • White Google+ Icon

©  2020  Ankit Rathi