Week 5: Implement the Basic Building Blocks of Random Forest Algorithm

[azmfaridee]

Parent Issue #3

As the per the initial proposal, start coding for the feature selection algorithm of regularized random forest, particular task would be implementation of the bootstrapping part as well as the helper functions. Relevant functions could be:

createBootStrappedSample()
selectAttributesToInclude()
getCategoryProbability()
getHighestCountCategory()
calculateEntropy()
calculateTreeErrorRate()
calculateAttributeImportance()
randomlyShuffleAttribute()

End of Week Deliverable:

Code segment that can create N times bootstrapped data than what was give at the beginning as well as other mentioned helper functions

Note:

• This is set as a guideline, we might end up with a completely different kind of function names and behavior than what has been described here
• Preparation of datasets must also be done, follow and update Issue #2 for that

[azmfaridee]

@kdiverson It's taking a little bit of time to warm up my C++ skill, I'm setting up the data structures, ran into a bit of problem with the vector of some classes as they are copied on reference, not copied by value. I'll take care of that in no time. Feel free to check out the code and let me know your observations.

[azmfaridee]

The functionality of these functions were implemented:

createBootStrappedSample()
selectAttributesToInclude()
getCategoryProbability()
getHighestCountCategory()
calculateEntropy()

The following functionality has not been implemented, I have postponed this to a following week, so that when the Trees are created, we calculate the error rate.

calculateTreeErrorRate()
calculateAttributeImportance()
randomlyShuffleAttribute()