Manikant92

Manikant92

Manikant92

#follow same process which was discussed before converting lists into dictionary then dataframe
new_dictionary = {'X': new_x, 'y':new_y}

new_df = pd.DataFrame(new_dictionary)
#split the data of 10 rows into 80% train set and 20% test set
new_X_train, new_X_test, new_y_train, new_y_test = train_test_split(new_df[['X']],new_df.y,test_size=0.2, random_state=5)
#fit to find the best fit
lr.fit(new_X_train, new_y_train)
#print slope
print(lr.coef_[0])

Manikant92

#convert into numpy arrays as scikit learn works on top of numpy array
new_x = np.array(new_x)
new_y = np.array(new_y)
#reshape x as we cannot use Rank 1 matrix in scikit learn
new_x = new_x.reshape(len(new_x),1)
#fit to determine slope and intercept by passing all input and output data
lr.fit(new_x,new_y)

print('Slope with Linear Regression SciKit Learn: ', lr.coef_[0])
print('Intercept with Linear Regression SciKit Learn: ',lr.intercept_)

Manikant92

#new input data which is represented in excel
new_x = [2,3,4,5,6,7,8,9,10,11]
#new output data which is represented in excel
new_y = [10,12,20,22,21,25,30,21,32,34]
#plot the graph
plt.scatter(new_x, new_y)
plt.ylabel('New Dependent Variable')
plt.xlabel('New Independent Variable')
plt.show()
#see plot Input and Output Scatter Plot before Best Fit

Manikant92

#slope -logic
m = lr.coef_[0]
#constant
c = lr.intercept_
#multiply m with all input data to form a regression line- linear equation to observe whether it is the best fit or not.
regression_line = [(m*i)+c for i in x]
#scatter plot whole x and y data.
plt.scatter(x,y,color='blue')
plt.plot(x,regression_line, color='red')
plt.ylabel('Dependent/Output Variable')

Manikant92

#logic is nothing but coefficient to inputs which gives our output
print(lr.coef_)

#output: [ 11.]

#intercept is the error/constant between input and output variables
print(lr.intercept_)

#output: -2.84217094304e-14 which is -0.0000000000000284217094304 in decimal notation

Manikant92

#predict by providing new input as -18
print(lr.predict(-18))

#output: [-198.]

#-18 * 11 = -198

#predict with 89 as input
print(lr.predict(89))

Manikant92

#import metrics module from sklear
from sklearn import metrics
print(metrics.r2_score(y_test, y_prediction)*100)

#output: 100.0

Manikant92

#X_test is our test set 5 rows-30% percent split. Pass our test data to predict method of algorithm.
y_prediction = lr.predict(X_test)
#print out the results of predictions done by our model.
print(y_prediction)

#output: [ 154.   33.  110.   55.   22.]