# Let's begin with machine learning.
# Machine learning is how computers learn from data.
 
print("With supervised learning, we know the answer for each example.")
print("With unsupervised learning, we only have the data itself.")

# Let's make a simple dataset for supervised learning.
fruits = ['apple', 'orange', 'banana', 'apple', 'orange']
labels = ['red', 'orange', 'yellow', 'red', 'orange']

print('Fruit:', fruits[0], '| Color:', labels[0])
print('Fruit:', fruits[2], '| Color:', labels[2])

# Here is some shopping data for unsupervised learning.
shopping = [5, 8, 6, 50, 54, 48]

print('Shopping amounts:', shopping)

# Let's compare supervised and unsupervised data.
print('Supervised has answers:')
print('fruits:', fruits)
print('labels:', labels)
print('Unsupervised just has data:')
print('shopping:', shopping)

# Let's import the library we need.
from sklearn.tree import DecisionTreeClassifier

# Here is a simple supervised learning example.
X = [[1], [2], [3], [4]]  # features, like fruit size
y = ['apple', 'apple', 'orange', 'orange']  # labels
 
clf = DecisionTreeClassifier()
clf.fit(X, y)  # train the model
 
print('Predict size 1:', clf.predict([[1]]))
print('Predict size 4:', clf.predict([[4]]))

# Import KMeans for clustering.
from sklearn.cluster import KMeans

# Now we find groups in our shopping data.
import numpy as np
shopping_array = np.array(shopping).reshape(-1, 1)

kmeans = KMeans(n_clusters=2, random_state=0)
kmeans.fit(shopping_array)

print('Predicted group for numbers:', kmeans.labels_)

# What happens if you give new data?
new_shopping = np.array([[7], [55]])
group = kmeans.predict(new_shopping)

print('Groups for new data points:', group)

# Challenge: classify animals by size (supervised learning)!
animals = [1, 10, 15]
labels = ['mouse', 'dog', 'sheep']

X_new = [[2], [12]]

animal_model = DecisionTreeClassifier()
animal_model.fit([[a] for a in animals], labels)

print('Classify size 2:', animal_model.predict([[2]]))
print('Classify size 12:', animal_model.predict([[12]]))

# Challenge: group people by height (unsupervised learning)!
heights = [145, 150, 160, 185, 190, 200]
heights_array = np.array(heights).reshape(-1, 1)
height_groups = KMeans(n_clusters=2, random_state=10)
height_groups.fit(heights_array)
print('Groups by height:', height_groups.labels_)
print('Which cluster is 170 in?', height_groups.predict([[170]]))

# Sample data: fruit size and colors
sizes = [[5], [7], [8], [11]]
colors = ['green', 'orange', 'yellow', 'red']

color_model = DecisionTreeClassifier()
color_model.fit(sizes, colors)

print('Predict for size 7:', color_model.predict([[7]]))
print('Predict for size 10:', color_model.predict([[10]]))

# Store shopping amounts for customers.
store_shopping = [10, 12, 13, 100, 110, 105]
shopping_data = np.array(store_shopping).reshape(-1, 1)

shopper_kmeans = KMeans(n_clusters=2, random_state=15)
shopper_kmeans.fit(shopping_data)
 
for i, amount in enumerate(store_shopping):
    print('Amount $' + str(amount), '-> Group', shopper_kmeans.labels_[i])
    

# Check a new customer's spending group.
amount_input = input('Enter new amount: ')
amount = float(amount_input)
predicted = shopper_kmeans.predict([[amount]])
print('This customer is in group:', predicted[0])

# Some useful tips and tricks!
print('You can try different models, like RandomForest or LogisticRegression!')
print('Try making your own datasets from hobbies or schoolwork.')

# Final challenge: Can you sort your favorite movies by rating?
movies = ['Movie A', 'Movie B', 'Movie C']
ratings = [7, 9, 6]
sorted_movies = [m for _, m in sorted(zip(ratings, movies), reverse=True)]
print('Your movies from highest to lowest rating:')
for movie in sorted_movies:
    print(movie)