My Background

I am a graduate student in Civil Engineering. For the analyses of road traffic data (vehicle trajectories as time series) I work with big data sets mostly about a million data points or more.
I started using R language when MS Excel could not open the big data files. Using basic statistics knowledge and R code I developed few algorithms to identify certain patterns in the data which worked for many applications. But I still lack serious programming skills in R.
Now, I am familiar with basic inferential statistics and R packages (plyr, dplyr, ggplot2, etc). Recently I came to know that Machine Learning algorithms also help in defining patterns in the data through supervised/ unsupervised learning and their application might improve the accuracy of prediction of certain 'behaviors' of drivers using the traffic data.


Having the basic knowledge of Statistics and R, I want to learn about the data science/ machine learning as a beginner. I know that some concepts in Stats. and ML overlap and that might bridge the gap in my learning of ML. Keeping my background in mind, what resources (books/ online courses) would you recommend me to start learning data science and apply it in my field?

  • $\begingroup$ Questions just looking for resources are considered off-topic for StackExchange. Maybe you can refine this into specific questions about specific tools. $\endgroup$
    – Sean Owen
    Nov 7, 2014 at 8:29

1 Answer 1


The best way to learn data science is through problem solving. I suggest you to head over to Kaggle and work through the for-knowledge problems.

To get a good start on Machine Learning problems, acclimate yourself with the tree package in R. This will help you understand how decision trees work, and building upon that, how random forests, gradient boosting machines and other sophisticated tree based algorithms work.

Then, there are SVMs and deep learning models.

To get an understanding of unsupervised learning problems, learn k-means and employ it for clustering.

Other general concepts/ ideas to understand are:

  1. cross-validation

  2. overfitting, regularization

  3. bias-variance trade off

  4. dimensionality reduction/ variable selection

  5. generalization error

  6. ensemble learning

For books, the most common recommendation to anyone who is familiar with statistics and wants to get into Machine Learning is "The Elements of Statistical Learning" by Hastie, Tibshirani, and Friedman.


Not the answer you're looking for? Browse other questions tagged or ask your own question.