Kiva

A quick data exploration of the Kiva dataset found on Kaggle. It's been sometime since I added a project. There are other things I hope to add soon, but I've been more leisurely about packaging them all and sharing

Forest Cover Type

A classification challenge from Kaggle that aims to predict the predomiant kind of tree cover in a section of a forest. Exploratory analysis is performed on the datset using visualization libraries Altair and Plotly. In addition an ensemble classifier model is used to predict the appropriate cover type.

PBS Newshour - Data Collection and Analysis

Collects all transcripts available on PBS Newshour's website. Topics includes:

• Data is retrieved through the Python webscraping library, Beautiful Soup.
• The data fetching is expediated through the Python threading library, Multiprocessing.
  • This is a much needed step as there are over 17 thousands unique URLs to scrap.
• Preprocessing cleans the text and determines each text's speaker. In addition:
  • Basic typos/errors like "PRESDIENT" or "\xa0" are fixed
  • Qualifiers like "MAYOR" or (D-CA) are removed
• The data is analysed using time series and word processing techniques

Mussel Watch - Exploratory Data Analysis

Looks through pollutation levels in our aquatic ecosystems. Topics include:

• Grouping pollutants and organisms into major categories (i.e. mussels and clams become bivalves).
• Looking at pollution levels for each substance over time.
• Viewing unique Mussel Watch sites across the United States.

2015 US Census Estimate - Exploratory Data Analysis

An exloration of the 2015 American Community Survey. Topics include:

• The unfortunate state of Puerto Rico in terms of poverty.
• Transporation differences across counties and states.
• Extremes in racial demographic populations.
• The strange non-binomial behavior of gender in certain counties.

Music to sheet music converter

Takes a .wav music file and convert it into sheet music through the following steps:

1. Segmentation: determines when each note starts in the file
2. Filtering: cutting off low/high frequencies, determining whether two notes are a chord or quick sequential notes
3. Pitch Detection: the pitch can be detected through Fast Fourier Transform and mapped onto a certain note
4. Filtering: determining whether a frequency hit is a partial or a note, determining whether a frequency hit is an independent note or residual from a few secconds earlier
5. Duration detction: finally determining how long a note lasts and it's location on a measure
6. Saving to Music XML file: the information gathered is organized, formated, and saved into a music XML file. NOTE: additional software is needed to view the sheet music XML. One good (free) choice is MuseScore

Introduction to Data Science

I'm not seasoned enough to teach Data Science, but I can curate what I think are the most important resources for learning Data Science. Hopefully this page keeps you and me organized as we learn more!

Kaggle Mini Projects