update this code like own work not copied from online tool

import numpy as np

import matplotlib.pyplot as plt

# Load the data from CSV

data $=$ np$.$genfromtxt$(\text{'}$Advertising$\_$N$200\_$p$3.$csv$\text{'},$ delimiter$=\text{'},\text{'},$ skip$\_$header$=1)$

# Separate the features and the target variable

X $=$ data$[$:$,$ :$-1]$ # TV$,$ Radio, Newspaper

y $=$ data$[$:$,-1]$ # Sales

# Add a column of ones for the intercept term

X $=$ np$.$column$\_$stack$(($np$.$ones$($X$.$shape$[0]),$ X$))$ # N x $($p$+1)$ design matrix

np$.$random.seed$(42)$ # For reproducibility

beta $=$ np$.$random.uniform$(-1,1,$ X$.$shape$[1])$ # p$+1$ parameters

def mini$\_$batch$\_$gradient$\_$descent$($X$,$ y$,$ beta, batch$\_$size$=10,$ learning$\_$rate$=2\mathrm{.}5$e$-6,$ iterations$=20000)$:

N $=$ X$.$shape$[0]$ # Number of observations

B $=$ N $//$ batch$\_$size # Number of batches per iteration

cost$\_$history $=[]$

beta$\_$history $=[]$

for itr in range$($iterations$)$:

# Shuffle data

permutation $=$ np$.$random.permutation$($N$)$

X$\_$shuffled $=$ X$[$permutation$]$

y$\_$shuffled $=$ y$[$permutation$]$

for b in range$($B$)$:

# Create mini$-$batch

start $=$ b $*$ batch$\_$size

end $=$ start $+$ batch$\_$size

X$\_$b $=$ X$\_$shuffled$[$start:end$]$

y$\_$b $=$ y$\_$shuffled$[$start:end$]$

# Compute the prediction

y$\_$pred $=$ np$.$dot$($X$\_$b$,$ beta$)$

# Compute the gradient

gradient $=-2*$ np$.$dot$($X$\_$b$.$T$,($y$\_$b $-$ y$\_$pred$))/$ batch$\_$size

# Update the parameters

beta $-=$ learning$\_$rate $*$ gradient

# Record cost and beta at each iteration

cost $=$ np$.$sum$(($y $-$ np$.$dot$($X$,$ beta$))**2)$

cost$\_$history.append$($cost$)$

beta$\_$history.append$($beta$.$copy$())$

return beta, cost$\_$history, beta$\_$history

final$\_$beta, cost$\_$history, beta$\_$history $=$ mini$\_$batch$\_$gradient$\_$descent$($X$,$ y$,$ beta$)$

beta$\_$history $=$ np$.$array$($beta$\_$history$)$

plt$.$figure$($figsize$=(10,6))$

for i in range$($beta$\_$history.shape$[1])$:

plt$.$plot$($beta$\_$history$[$:$,$ i$],$ label$=$f'Beta $\{$i$\}\text{'})$

plt$.$xlabel$(\text{'}$Iteration$\text{'})$

plt$.$ylabel$(\text{'}$Beta Coefficients'$)$

plt$.$title$(\text{'}$Effect of Iterations on Beta Coefficients'$)$

plt$.$legend$()$

plt$.$show$()$

plt$.$figure$($figsize$=(10,6))$

plt$.$plot$($cost$\_$history$)$

plt$.$xlabel$(\text{'}$Iteration$\text{'})$

plt$.$ylabel$(\text{'}$Cost$\text{'})$

plt$.$title$(\text{'}$Effect of Iterations on Cost'$)$

plt$.$show$()$

print$("$Best$-$fit model parameters:", final$\_$beta$)$

mse $=$ np$.$mean$(($y $-$ np$.$dot$($X$,$ final$\_$beta$))**2)$

print$("$Mean Squared Error on training set:", mse$)$