Add PCA from Scratch

Principal Component Analysis (Dimensionality Reduction)/PCA.py

@@ -0,0 +1,24 @@
+import numpy as np 
+
+class PCA:
+	def __init__(self, reduced_to_k):
+		self.reduced_to_k = reduced_to_k
+		self.Mean = None 
+		self.final_components = None
+
+	def fit(self, X):
+		self.Mean = np.mean(X, axis = 0)
+		X = X - self.Mean
+		covariance = np.cov(X.T)
+		evalues, evectors = np.linalg.eig(covariance)
+		evectors = evectors.T
+		indexes_maxi_k = np.argsort(evalues)[ : : -1]
+		evectors = evector[indexes_maxi_k]
+		self.final_components = evectors[0 : self.reduced_to_k]
+
+	def transform(self, X):
+		X = X - self.Mean
+		FINAL_COMPONENT_MATRIX = np.dot(X, self.final_components.T)
+		return FINAL_COMPONENT_MATRIX
+
+