Information Cascade and Election

What is Information Cascade?

Information cascade is referring to the effect that people's decisions being affected by the choices of a large group of people, particular when people make decisions sequentially.

For example, when people decide which products to buy, the best selling product often attracts more customer. The agent followed the decisions of other customers.

Information cascade occurs when:

1. Agents make decisions sequentially
2. Agents make decisions rationally based on the information they have
3. Agents do not have access to the private information of others
4. A limited action space exists (e.g. an adopt/reject decision)

Information Cascade in Social Media

From the lecture, an experiment in Facebook was introduced to test the effect of information cascade in social media. The result showed the effect did happen.

Actually not only did it happen in experiments, it also affected the voting rate of 2012 US president election.

In US president election 2002, Facebook launch a new feature called "I'm Voting". When a user pressed the button, his/her friends will see the "I'm voting" button appeared on the user's top newsfeed.

Election fulfilled the four condition required for information cascade.

1. Voters vote sequentially
2. Voters make their decisions (vote or not? who to vote?) rationally
3. Voters do not have the full picture of the decisions made by others
4. The action space is limited

Thus, in theory, information cascade can happen. In fact, study found out it boosted the turnout rate by 2.2%, proving information cascade did happen.

So what?

The experiment and the real life case showed the power of information cascade in social media. Its influential power far exceeds traditional media. I can predict that the different parties will shift more and more attentions from traditional media to social media. A lot of opportunities are awaiting to be dug out. 

Overfitting in Supervised Learning

Supervised Learning

In Lecture 4, the idea of supervised classification was introduced.

Supervised classification is a way to solve the classification problem by first training the classifier with data of known classification. Then, the classifier can do classification based on the known data set. For example, in assignment we trained the classifier with two pieces of data, namely "health.txt" and "business.txt". Then it can classify the new input words with them.

(Figure from http://www.nltk.org/book/ch06.html)

Overfitting

During the training, we often focus on the accuracy of the classifier. But being too accurate with the training data may introduce another problem - overfitting.

Overfitting means "a statistical model describes random error or noise instead of the underlying relationship" (source). A set of training data likely contains some noise and errors that are not related to the classification. For example, in our assignment 2, both health.txt and business.txt contains words that are neither health related nor business related, such as "slow", "system". If we train a classifier by attempting to perfectly fit the training data, those unrelated words may be classified wrongly. Overfitting occurs.

Bootstrap aggregating

In order to solve overfitting, a method called boostrap aggregating, or bagging.

Using bagging, a number of new training sets can be generated from the original training set. It is done by sampling the original training set with replacement - i.e. each word can be selected more than one time. Usually hundreds or thousands of boostrap samples are used.

After getting the samples, predictions will be by voting among all the samples. The result which gets the majority vote will be the prediction. By considering a large number of boostrap sample, bagging reduced the effort of overfitting.

An improvement over bootstrap aggregating is Random Forest. It can improve the prediction result if one or a few features are a strong predictor.