search

DeltaAnalytics / machine_learning_for_good

Machine learning fundamentals lesson in interactive notebooks
http://www.deltanalytics.org/
Creative Commons Attribution 4.0 International
174 stars 123 forks source link

Added Laplace smoothing parameter tuning #59

Closed amankhullar closed 4 years ago

amankhullar commented 4 years ago

Tuned laplace smoothing parameter to the Naive Bayes classifier in module 7.

I experimented with 3 different values 0.01, 1 (default) and 2 and have noted that both the parameters 0.01 and 2 improve upon the default value of 1. However there is much more significant improvement when the parameter value is closer to 0. This suggests that not a lot of smoothing is required. However, would be happy to discuss any other inferences we can draw.

brianspiering commented 4 years ago

I got a chance to look at your work. It is a nice addition.

I had to manually copy n' paste your cells into the master branch to avoid a messy merge conflict.

Your work is now on the master branch.

I'm closing this PR without merging it.

Please keep contributing.

amankhullar commented 4 years ago

Thank you, I am glad to be able to contribute, will try to avoid messy merge conflicts the next time. Sure, would be happy to keep contributing!