predict method

Lesson 14 of 19 in Coddy's Introduction to Machine Learning course.

Challenge

Easy

The predict assigns (or predicts) a class to a new data point using the following formula:

temperature	mood	motivation	windy
low	happy	high	yes

p_yes = probability_features["temperature"]["low"]["yes"] *
	probability_features["mood"]["happy"]["yes"] *
	probability_features["motivation"]["high"]["yes"] *
	probability_features["windy"]["yes"]["yes"] *
	probability_target["yes"]

Previous lesson we wrote the get_classes_probability.

Now in the predict method we need to return the output of get_classes_probability of the new data point.

Try it yourself

class NaiveBayes:
    def __init__(self):
        pass
    
    def get_classes_probability(self, X):
        # X holds a dataframe with a single row
        # use self.classes that was saved in fit method
        # write your code below
        classes_probabilities = {}
        for cls in self.classes:
            classes_probabilities[cls] = 1
            for column in X.columns:
                for distinct_value in X[column].unique():
                    classes_probabilities[cls] *= self.probability_features[column][distinct_value][cls]
            classes_probabilities[cls] *= self.probability_target[cls]
        # --------------------------
        return classes_probabilities
    
    def fit(self, X_train, y_train):
        self.probability_features = {}
        self.probability_target = {}
        self.classes = y_train.unique()
        target_variable = "target"
        df = X_train.copy()
        df[target_variable] = y_train
        # Copy your code from previous lesson
        # Don't forget to Add self. to probability_features and probability_target
        distinct_target_values = df[target_variable].unique()
        for column in df.columns:
            if column == target_variable:
                continue
            self.probability_features[column] = {}
            distinct_values = df[column].unique()
            for value in distinct_values:
                self.probability_features[column][value] = {}
                for target_value in distinct_target_values: 
                    total = len(df[(df[column] == value) & (df[target_variable] == target_value)][target_variable])
                    count = len(df[(df[column] == value)])
                    self.probability_features[column][value][target_value] = total / count

        for target_value in distinct_target_values: 
            total = len(df[(df[target_variable] == target_value)][target_variable])
            self.probability_target[target_value] = total / len(df[target_variable])

        
    
    def predict(self, X_test):
        pass