12. Assignment 12: Fake News#

12.1. Quick Facts#

Submit by the first feedback deadline to get feedback and then continue revising your work. After the final due date that feedback will be final.

12.2. Evaluation#

Table 12.1 use text to predict fake from real news#

task

skill

transform text data to a format compatible with ML

representation (2)

plan to solve a real world problem using the tools from class

workflow (2)

usse EDA techniques to interpret the experimental results

summarize (2), visualize (2)

determine the best model for this task

compare (2)

determine the best parameter settings for this task

optimize (2)

12.3. Instructions#

Use the dataset in the assignment template repo to answer the following questions. The data includes variables:

  • ‘text’: contents of an article

  • ‘label’: whether it is real or fake news

  • ‘title’: title of the article

  1. Is the text or the title of an article more predictive of whether it is real or fake?

  2. Are titles of real or fake news more similar to one another?

Consider what difference you can have in how you represent the data and how that might impact your model performance in order. Use summary statistics and visualizations appropriately in order to explain your results.

Hint

The data set contains a large number of articles (takes a long time to train), you can downsample this to something like a 1,000 articles or so in order to speed up training and evaluation (hint: use shuffle).