Merge pull request #119 from jeffin07/master

Multivariable Linear Regression

Author: master-fury

Machine_Learning/MultivariateLinearRegression.py

@@ -0,0 +1,64 @@
+import numpy as np 
+import pandas as pd
+import matplotlib.pyplot as plt  
+
+class LinearRegression:
+	def fit(self, x, y, learning_rate):
+		self.n_weights = np.zeros(x.shape[1])
+		self.learning_rate = learning_rate
+		self.loss_=[]
+		print("Initial cost {} ".format(model.cost_function(x, y, self.n_weights)))
+		model.gradient(x, y, self.n_weights, 10000)
+		print("Final cost {} ".format(model.cost_function(x, y, self.n_weights)))
+		return self.n_weights
+
+
+	def cost_function(self, x, y, n_weights):
+		n = len(y)
+		cost = np.sum((x.dot(self.n_weights.T) - y) ** 2) / (2 * n)
+		return cost
+	
+
+
+	def gradient(self, x, y, n_weights, epochs):
+		m = len(y)
+		for i in range(epochs):
+			h = x.dot(n_weights.T)
+			loss = h - y
+			change=(x.T.dot(loss) / m) * self.learning_rate
+			self.n_weights -= change
+			self.loss_.append(model.cost_function(x, y, self.n_weights))
+			if i % 10 == 0:
+				print("Loss of {}th epoch is {} ".format(i , model.cost_function(x, y, self.n_weights)))
+		return self.n_weights
+
+	def predict(self, x):
+		x=np.insert(x, 0 ,1)
+		print(x.T.dot(self.n_weights))
+
+
+
+	def plot(self):
+		plt.plot(self.loss_)
+		plt.xlabel("Epochs")
+		plt.ylabel("Loss")
+		plt.show()
+
+
+
+if __name__ == "__main__":
+	#Importing data and some preprocessing
+	data = pd.read_csv('student.csv')
+	data["one"] = [1 for i in data["Math"]]
+	math = data["Math"]
+	write = data["Writing"]
+	read = data["Reading"]
+	one = data["one"]
+	x = np.array([one,math,read]).T
+	y = np.array(write)
+	learning_rate = 0.0001
+	model = LinearRegression()
+	model.fit(x, y, learning_rate)
+	print("Plotting loss")
+	model.plot()
+	model.predict(np.array([45,48]))