Machine Learning with Text

Standard

I started learning how to do machine learning on text data. I made a Python program with a small training set using the sklearn module to predict if a message is important or not. Since it is a small training set the results are not that accurate. This could be used for an app that learns what texts are considered important over time and marks them. Then the user manually decides if a text is actually important or not and the app learns what texts are considered important from this.

link to download the Python code

Here is an image of the output:

Screen Shot 2017-07-07 at 4.18.44 PM.png

 

 

Advertisements

Habitable Exoplanets

Standard

I developed a machine learning Python program that uses a Support Vector Machine classifier from the sklearn module to predict if an exoplanet is in the habitable zone of its star system (the zone where liquid water can exist). The features of the planet required by the classifier to make the prediction are the orbital period of planet in days, number of planets in the system, and number of stars in the system.

I used the astroquery module to obtain the data from the Open Exoplanet Catalog

Link to download Python code

Here are some pictures of the program:

Screen Shot 2017-06-09 at 1.38.30 PM.png

Predicting Intel Revenue

Standard

I wrote a machine learning Python program using the sklearn module that predicts the revenue for Intel Corporation. I got the data for the training set from a website called Quandl that has a wide variety of financial data. I used Ridge regression with a 3rd degree polynomial for my machine learning algorithm.

Link to download Python code

Here is a picture of the output of my program:

Screen Shot 2017-06-06 at 3.58.49 PM.png

Classifying Hydrocarbons

Standard

I developed a machine learning Python program using the sklearn module that classifies a simple single-chain hydrocarbon  into categories based on the type of bonds it has given the number of carbons and hydrogens (alkanes have single bonds, alkenes have double bonds and alkynes have triple bonds).

Although there are easy ways to tell what type of bonds a simple single-chain hydrocarbon has given the number of carbons and hydrogens, I wanted to use a training set and a Linear SVM(Linear Support Vector Machine) to classify them in order to introduce myself to machine learning.

Link to download Python code

Here is a picture of the output of my program:
Screen Shot 2017-06-04 at 4.18.25 PM.png