Simple machine learning projects to learn machine learning

projects
machine_learning

#1

I am relatively new to machine learning - have worked on a few regression models, decision trees and clustering in the past. I want to now learn the broader spectrum of tools for various kinds of machine learning, but want to do it in hands on manner.

To be clear, I am not looking for advice like - go and do the machine learning course on Coursera of edX. I am looking for specific inputs like:

  • Build a image classification model on ABC dataset - x is an acceptable score.

I am already working on a few knowledge problems from Kaggle - let me know if you know any thing of this type further.


Machine Learning repositories
#2

@oliver You are definitely taking the right approach. Working hands on a few projects is definitely the best approach to learn machine learning.

Here is a list of projects, you may find useful:

  • MNIST dataset for image recognition - probably the best starting point for practicing deep learning and neural networks. http://yann.lecun.com/exdb/mnist/
  • Datasets from UCI machine learning repository: http://archive.ics.uci.edu/ml/. There are a lot of problems for classification, regression, and clustering, most of them would be used on an average computer.
  • Enron Email dataset - One of the common dataset for classification problem. http://www.cs.cmu.edu/~enron/
  • You can look at the assignments from CS109 from Harvard - this should be a good place to get hands dirty, especially so, if you are using Python

Depending on the tool you might be using, it would come pre-loaded with datasets for learning. For example scikit learn comes with datasets for building models - you can check them out as well