Skip to content

Latest commit

 

History

History
52 lines (30 loc) · 2.19 KB

File metadata and controls

52 lines (30 loc) · 2.19 KB

Exploratory Data Analysis

The module can be seen as a collection of standalone notebooks, each one covering a more niche topic in data science. The notebooks provide in-depth explorations of specific analysis topics that students are generally expected to "know beforehand" or pick up on the fly.

The first notebook covers Exploratory Data Analysis, a technique used to quickly gain a broad understanding of a (typically unfamiliar) dataset as a means to inform and guide future analyses.

The second notebook explains and investigates four common preprocessing techniques used in machine learning (Minmax, Standard, Robust, and Quantile scaling).

Learning goals

After completing this module, students should feel comfortable:

  • performing short, spontaneous analyses to familiarize themselves with a new dataset
  • choosing a preprocessing technique for a given dataset before running a more intensive machine learning analysis on it
  • using built-in functions from numpy and scipy to preprocess data before running a more intensive machine learning analysis on it

Prerequisites

Students are expected to be familiar with python, numpy, and matplotlib.

What's included in this module?

In-class notebooks

Two in-class notebooks WILL guide students through various preliminary topics in data analysis. The notebooks will cover:

  • Exploratory data analysis (01_dataset_cleaning.ipynb)
  • Preprocessing techniques for neural networks (02_preprocessing_techniques.ipynb)

Estimated time to complete: 30-45 minutes per notebook

Conceptual Quiz Questions

N/A

Homework Assignment

N/A

Datasets

Two datasets are provided to accompany the notebook 01_dataset_cleaning.ipynb.

What courses might this module accompany?

This module would be well-suited to accompany a general course in data science techniques for the physical sciences (or a specific subfield of a physical science). The first notebook in particular would be very apropriate as a first homework assignment for any laboratory course.


TODO

  • Possibly adding in a quiz module
  • Possibly adding in a HW module
  • Remove some of the code to students have something to do