Assignment3:

Q) What parameters/options are we allowed to change in our effort to improve the accuracy rate?

A) All the available parameters, as long as you meet the assignment requirements.

Q) Do we have to stick to 10-fold? etc. Does higher percentage accuracy equal higher assignment grade? 

A) No, please check the Grading Criteria section in the project description: https://coursys.sfu.ca/2021sp-cmpt-310-d1/pages/project

Q) What is predictive accuracy? Where do we get this value? In the assignment description, 

A) Generally speaking, predictive accuracy is the ratio of the number of correct predictions out of all predictions made, often presented as a percentage (100% is the best an algorithm can achieve). For this assignment, our primary evaluation is 10-fold cross-validation.

Q) Will changing the pre-processed input make a difference? Are we allowed to try different filters for the pre-processing step? Do we have to explain what the filters are doing in this step in our report? 

A) Yes, you're allowed to explore different preprocessing options. You don't need to deeply explain the utilized approach, but in your report, you need to mention the applied stages.

Q) Can we use different trees other than J48? 

A) Yes, if you're confident how they work

Q) Should we submit our stopwords list, or other extra information that we use?

A) Yes, please. Or give us a pointer to download them.

Q) How much to report? 

A) The project report should be at most 2 pages using no smaller than 11pt font (excluding references).  You do not have to use both pages, conciseness is good.

Q) How to chain the filters in WEKA?

A) Using the chaining filters option. Please checkout this: http://jmgomezhidalgo.blogspot.com/2013/02/text-mining-in-weka-revisited-selecting.html