Hello! Today I am going to share some basic information about machine learning algorithms that I learned while doing research on the the effects of machine learning methods on the relationship between social media and substance use.

What are machine learning algorithms?

A newer tool in data analysis is the use of machine learning algorithms to analyze data and make predictions based on the data collected. There are different types of machine learning models including support vector machines, regression, deep learning models, and maximum entropy text classifiers.

When data is collected, there are often two or more different explanatory variables in addition to the response variable for the data set. Machine learning models are able to take into account all of the different factors and variables for each piece of data and make predictions from that data.

Common machine learning algorthims:

Support vector machines create digital vectors for each piece of data and perform analyses to classify the data.

Regression is similar, but it uses different mathematics and statistics to analyze data. There are also different types of regression including linear regression and multiple linear regression.

Linear regression analyzes the linear relationships between two variables in order to predict the response of one variable from the value of another.

Multiple linear regression does the same thing, but with more than two variables.

Additionally, deep learning models are another way to analyze data. Deep learning models create networks and matrices for a data set and then are able to perform different types of analyses for the data. Deep learning models build connections and analyze the relationships from a data set in order to classify or make predictions about the data.

The last machine learning method discussed in this post is maximum entropy text classifiers. Maximum entropy text classifiers for text analyze text data through an importance method and determine priority through correlations of the most important words versus the least important words.

What does it mean if an algorithm is supervised?

Another important aspect of machine learning algorithms is that the algorithm can be supervised or unsupervised.

Supervision includes creating a training data set and a data set for real analysis. The training data set includes data collected and the expected or correct classification. Then, the training data set it used to teach the algorithm how to classify the collected data. Next, the machine learning algorithm is able to look at the real data and classify it.

Unsupervised algorithms do not have a training data set and instead are used directly on the collected data set.

The type of supervision depends on the type of data analyzed and the outcome desired. Some machine learning algorithms are supervised while others are not.