# To start, let us suppress warnings for a smoother learning experience.
import warnings
import numpy as np
np.random.seed(42)
warnings.filterwarnings("ignore")

# Data setup (Iris Dataset)
import pandas as pd
url = 'https://raw.githubusercontent.com/uiuc-cse/data-fa14/gh-pages/data/iris.csv'
df = pd.read_csv(url)
print(df.shape)
print(df.head(3))

# Let us look at our data types and check for missing values.
print(df.dtypes)
print(df.isnull().sum())

# PCA needs only the numeric features. Let us separate them now.
X = df.drop('species', axis=1)
print(X.head(3))

# Normalizing features: PCA works best when features are on the same scale.
from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)
print(X_scaled[:3])

# Let us apply PCA to our iris data. We will use only two principal components for visualization.
from sklearn.decomposition import PCA
pca = PCA(n_components=2)
X_pca = pca.fit_transform(X_scaled)
print('Shape after PCA:', X_pca.shape)
print(X_pca[:3])

# Let us add species labels back for plotting.
import numpy as np
pca_df = pd.DataFrame(X_pca, columns=['PC1', 'PC2'])
pca_df['species'] = df['species']
print(pca_df.head(3))

# Let us plot the PCA result to see if the species separate well.
import matplotlib.pyplot as plt
for s in pca_df['species'].unique():
    plt.scatter(pca_df[pca_df['species']==s]['PC1'], pca_df[pca_df['species']==s]['PC2'], label=s)
plt.xlabel('PC1')
plt.ylabel('PC2')
plt.legend()
plt.title('Iris dataset visualized using PCA')
plt.show()

# Let us see how much information each principal component keeps.
pca_full = PCA().fit(X_scaled)
explained_var = pca_full.explained_variance_ratio_
print('Variance explained by each component:', explained_var)
print('Total explained by first two:', explained_var[:2].sum())

# Let us try running PCA on the Titanic dataset for more practice.
url = 'https://raw.githubusercontent.com/datasciencedojo/datasets/master/titanic.csv'
titanic_df = pd.read_csv(url)
print(titanic_df.shape)
print(titanic_df.head(3))

# Let us preprocess Titanic: fill missing numbers and drop text clues.
num_cols = titanic_df.select_dtypes(include='number').columns
filled_df = titanic_df[num_cols].fillna(titanic_df[num_cols].mean())
print(filled_df.head(3))

# Now, let us scale and apply PCA to the Titanic numbers.
scaled_fill = scaler.fit_transform(filled_df)
titanic_pca = PCA(n_components=2)
titanic_pcs = titanic_pca.fit_transform(scaled_fill)
print('Shape:', titanic_pcs.shape)
print(titanic_pcs[:3])

# PRACTICE PROMPT: Try running this with three or more PCA components.

n = int(input("How many components would you like to use? (2-4): "))
titanic_pca_n = PCA(n_components=n)
titanic_pcs_n = titanic_pca_n.fit_transform(scaled_fill)
print(f'Titanic PCA with {n} components shape:', titanic_pcs_n.shape)
 
 

# Mini-project: Visualize Iris dataset in 3D using the three first PCA components.
pca_3d = PCA(n_components=3)
iris_pca3 = pca_3d.fit_transform(X_scaled)
from mpl_toolkits.mplot3d import Axes3D
fig = plt.figure(figsize=(8,6))
ax = fig.add_subplot(111, projection='3d')
colors = {'setosa':'r', 'versicolor':'g', 'virginica':'b'}
for s in pca_df['species'].unique():
    label_mask = (df['species'] == s)
    ax.scatter(iris_pca3[label_mask,0], iris_pca3[label_mask,1], iris_pca3[label_mask,2],
               label=s, c=colors[s], s=50)
ax.set_xlabel('PC1')
ax.set_ylabel('PC2')
ax.set_zlabel('PC3')
plt.legend()
plt.title('Iris in 3D PCA space')
plt.show()

# Best Practices: Always scale before PCA and check explained variance.
# Try different numbers of components and look at the plot or explained variance.