Reviewer: @akamlani
Master Check List:
- Good job on the wrangling of the 'iris.csv' into a format that can be used with the sklearn package.
- Multiple ways in terms of getting the optimal k in terms of maximum accuracy, getting maximum number from tuple via index works well!
- Good use of subplots to show the entire range of k vs a 'zoomed-in' sample size of k for the classifier accuracy. I agree, it is easier to tell the local maxima in the 'zoomed-in' version of the scatter plot.
- Curious, did you try the other numbers of folds for the bonus question, instead of the samples that were given? For example what happens when the number of folds is chosen quite large? From the set of folds given, yes it doesn't look like their exists an optimal number of folds. However perhaps we could look at other samples of fold size.
@tomharel
/cc @ghego, @craigsakuma, @kebaler
Reviewer: @akamlani
Master Check List:
- Good job on the wrangling of the 'iris.csv' into a format that can be used with the sklearn package. - Multiple ways in terms of getting the optimal k in terms of maximum accuracy, getting maximum number from tuple via index works well! - Good use of subplots to show the entire range of k vs a 'zoomed-in' sample size of k for the classifier accuracy. I agree, it is easier to tell the local maxima in the 'zoomed-in' version of the scatter plot. - Curious, did you try the other numbers of folds for the bonus question, instead of the samples that were given? For example what happens when the number of folds is chosen quite large? From the set of folds given, yes it doesn't look like their exists an optimal number of folds. However perhaps we could look at other samples of fold size.
@tomharel
/cc @ghego, @craigsakuma, @kebaler