Exploratory Data Analysis with Python
1. Learning how to find general details about a dataset before jumping on to the machine learning / big data bandwagon. (I’m calling these things ‘bandwagon’ because they are incredibly powerful, and in many cases, the application might not warrant a full scale use of such tools.)
2. Learning to decide which tools are best for taking apart a big dataset.
3. Understanding why getting a general feel of the data is necessary before thinking up models to analyze that data.
So we have a large data file. We might not know what to do with it. We most probably are looking for patterns and trends. With a multitude of data analysis tools and algorithms at our disposal, we are often left wondering as to what’s the right thing to ask of the data.
Exploratory data analysis is a field which offers tools and algorithms for the most broad, general look at a piece of data. It is after performing this sort of a global analysis on the data that we can go ahead and think about building a model to describe the data. This tutorial offers insights into the prerequisites for building such models, and having gained those, what all one could do with the model.
The tutorial will seek to answer questions like:
- What’s the best way to cluster / classify a given dataset?
- What does the data ‘look’ like?
- How has the dataset evolved over time?
- How do I know that I have inferred all I can from the dataset?
- I see some peculiar trends in the dataset. What might have caused these?
- Do all these questions motivate a good machine learning problem?
- A basic knowledge of Python. (Knowing how to use the numpy.ndarray object will be a plus.)
- Basic probability and statistics.
- Basic knowledge of popular data formats.
- File handling and I/O.
- Preferably a laptop with the free version of the Enthought Python Distribution installed. (One step solution to everything you’ll need in Python for scientific computing.)
I am an electrical engineering undergrad at VIIT, Pune. I’ve been working as a research assistant in the fields of machine learning and signal processing. I am currently working as an intern at Enthought, Inc, where I work on data analysis and visualization. I also contribute code and documentation to the SciPy project.