Predict articles that will generate negative impact(reactions on Facebook)
For the current assignment you are to read the dataset understand it and predict which articles will generate negative impact which could be based on the reactions that the user gives on the provided Facebook dataset. You need to analyze the dataset and provide some good insight of it like the number of reactions used, which topic is the most talked about topic and which topics generate negative impact through different visualization techniques. Your code must have all data mining stages and any data mining task https://en.wikipedia.org/wiki/Data_mining#Process.
The link provided can be used to see what data mining stages are to be included and select any data mining task. You must also show the accuracy of your model by printing out the scores.
You need to add comment box as to why you have selected the method. The code that is
submitted must be thoroughly documented. Any submission that does not have enough
documentation or documentation that is unclear will not get any points.
Suggestions for the Project:
1. You can remove any rows in the dataset that does not have any reactions and just keep the rows that have some numeric value.
2. Yu can also divide the reaction as positive and negative where negative contains too sad and to angry all the other reactions can be grouped under the positive side.
3. You can use various methods for the modelling to show how the accuracy improves from one model to other and in the end, you could plot a graph for the accuracy for all the models.
4. You could classify if the article will have negative impact based on the fact that which articles have large amount of to sad and to angry reactions. You are free to use any other method also.
5. You can also display a word cloud which represents the topics that have received negative reactions.
For reference you could also go through different papers some of them are given here:
1. https://arxiv.org/pdf/1905.10975.pdf
2.https://towardsdatascience.com/project-topher-facebook-page-reaction-prediction-program-95bcf4916892
The final submission should have all the code documented with a video recording.