$4.$ We use Machine Learning to derive insights from the data $($descriptive analysis and

modeling$)$ AND to infer or predict things about new data $($predictive modeling$).$ You will

now generate multiple predictive models, compare their performance, and use your

models to predict which type of iris is most likely for new observations.

a$.$ First, we need to split out data into a subset used for training our models and a

subset use for validating $($comparing$)$ our models. The method for this is

train$\_$test$\_$split. A common syntax for doing a random $80\%/20\%$ split of our

independent variables X $($see above$)$ and dependent variable Y $($class$$ in this

dataset$)$ is:

X$\_$train, X$\_$valid, Y$\_$train, Y$\_$valid $=$ train$\_$test$\_$split$($X$,$ y$,$ test$\_$size$=0\mathrm{.}20,$ random$\_$state$=1)$

b$.$ Once you have split your dataset into train and validation sets, you will need to build

a model that will predict the species based on our four input features. First build a

$$Decision Tree Classifier$$ in your notebook $$ train than classifier, use the model to

predict the iris type for each of the entries in the validation dataset, and compute

the accuracy of the prediction:

# Create a decision tree classifier

clf $=$ DecisionTreeClassifier$()$

# Train the model on the training data

clf$.$fit$($X$\_$train, y$\_$train$)$

# Make predictions on the test data

y$\_$pred $=$ clf$.$predict$($X$\_$valid$)$

# Calculate accuracy

accuracy $=$ accuracy$\_$score$($y$\_$valid, y$\_$pred$)$

printf$($Accuracy: $\{$accuracy$*100$:$\mathrm{.}2$f$\}\%$