Hello! Help with python, please!

For numerical features, impelement a class that fills missing values in each feature with its mean value.

Implement a fit method that takes as input feature matrix X and computes and saves mean value of each feature $($assuming that all the features are numerical$).$ The feature matrix can be either pd$.$DataFrame, or np$.$ndarray.

Implement a transform method that takes as input feature matrix X and replaces NaN values in each feature with a corresponding mean value and returns a transformed feature matrix.

from sklearn.base import TransformerMixin, BaseEstimator

class MeanImputer$($BaseEstimator$,$ TransformerMixin$)$:

def fit$($self$,$ X$,$ y$=$None$)$:

$"""$

Compute and save mean value of each feature in the feature matrix.

Parameters

$----------$

X : pd$.$DataFrame or numpy.ndarray of shape $($n$\_$samples, n$\_$features$)$

Feature matrix.

y : pd$.$Series or numpy.ndarray of shape $($n$\_$samples,$)($default : None$)$

Target values. Optional.

Returns

$-------$

self : returns an instance of self.

$"""$

self.column$\_$mean $=[]$

for i in range$($X$.$shape$[1])$:

if type$($X$)$ is not np$.$ndarray:

X$\_$i $=$ X$.$values$[$:$,$ i$]$

else:

X$\_$i $=$ X$[$:$,$ i$]$

# calculate mean for each column and add to the list $`$self$.$column$\_$mean$`$

# YOUR CODE HERE

raise NotImplementedError$()$

return self

def transform$($self$,$ X$)$:

$"""$

Fills missing values in each feature with a corresponding mean value.

Parameters

$----------$

X : pd$.$DataFrame or numpy.ndarray of shape $($n$\_$samples, n$\_$features$)$

Feature matrix.

Returns

$-------$

X$\_$transformed : array$-$like of shape $($n$\_$samples, n$\_$features$)$

Transformed feature matrix.

$"""$

X$\_$transformed $=[]$

for i in range$($len$($self$.$column$\_$mean$))$:

if type$($X$)$ is not np$.$ndarray:

X$\_$i $=$ X$.$values$[$:$,$ i$]$

else:

X$\_$i $=$ X$[$:$,$ i$]$

# Fill missing values using mean values

# YOUR CODE HERE

raise NotImplementedError$()$

X$\_$transformed $=$ np$.$column$\_$stack$($X$\_$transformed$)$

return X$\_$transformed

# TEST MeanImputer class

A $=$ np$.$array$([$

$[0,$ np$.$nan, $3],$

$[$np$.$nan, $6,7],$

$[$np$.$nan, $8,$ np$.$nan$],$

$])$

mean$\_$imp $=$ MeanImputer$()$

mean$\_$imp.fit$($A$)$

print$($mean$\_$imp.column$\_$mean$)$

assert mean$\_$imp.column$\_$mean $==[0\mathrm{.}0,7\mathrm{.}0,5\mathrm{.}0],$ "Computed mean values are incorrect."

mean$\_$imp.fit$($pd$.$DataFrame$($A$))$

A$\_$transformed $=$ mean$\_$imp.transform$($pd$.$DataFrame$($A$))$

assert ~np$.$isnan$($A$\_$transformed$).$any$(),$ "Transformed feature matrix still contains NaNs."

assert np$.$allclose$($A$\_$transformed,

$[[0.,7.,3.],$

$[0.,6.,7.],$

$[0.,8.,5.]]),$ "Filled values are incorrect."