I need help with task 1, can anyone help me solve it?

Task 1: Use PCA from previous assignment to convert data from 4D to 2D. If you did not manage to implement PCA, you can simply use first 2 dimensions of the database. (4 points if PCA is used, otherwise 1 point)

Task 2a: Implement kNN classification algorithm and apply it to classify iris database. (2 points)

Task 2b: Try different values of k = {1, 2, 3, 4, 5} and print prediction accuracy for validation set. Please select optimal value of k and justify your selection. (3 points)

Task 3: Implement random forest classification and apply it to classify iris database. Use any reasonable parameters you want. (3 points)

Taks 4: Visualize results of kNN and random forests (4 points)

# Load packages as usual

%matplotlib inline

import numpy as np

import pandas as pd

import matplotlib.pyplot as plt

import math

import random

import matplotlib.cm as cm

import numpy.matlib

from matplotlib.colors import ListedColormap

from sklearn.neighbors import KNeighborsClassifier

from sklearn.ensemble import RandomForestClassifier

# Manipulating figure sizes

import matplotlib

matplotlib.rcParams['figure.figsize'] = (15,7)

matplotlib.rc('font', size=15)

matplotlib.rc('axes', titlesize=15)

def __read(fileName, pTrainSamples = 0.6, pValidSamples = 0.2):

emp_df = pd.read_csv(fileName)

values = emp_df.values

values = emp_df.values.astype(np.float)

nTrainSamples = int(values.shape[0] * pTrainSamples)

nValidSamples = int(values.shape[0] * pValidSamples)

trainingFeatures = values[0:nTrainSamples, 0:-1]

trainingLabels = values[0:nTrainSamples, -1]

validationFeatures = values[nTrainSamples:nTrainSamples + nValidSamples, 0:-1]

validationLabels = values[nTrainSamples:nTrainSamples + nValidSamples, -1]

testingFeatures = values[nTrainSamples + nValidSamples:, 0:-1]

testingLabels = values[nTrainSamples + nValidSamples:, -1]

return trainingFeatures.astype(np.float), trainingLabels.astype(np.int), \

validationFeatures.astype(np.float), validationLabels.astype(np.int), \

testingFeatures.astype(np.float), testingLabels.astype(np.int)

trainingFeatures, trainingLabels, validationFeatures, validationLabels, testingFeatures, testingLabels = __read('C:/Users/me/iris_new.csv')

# Converting data to two dimensions using PCA

# The conversion to 2D from 4D input is needed to simplify visualization of the results. Please generate PCA on the training, and then convert training and testing data from 4D to 2D.

def __PCA(data):

# Extract data dimensions

d, N = data.shape

# First, center the data

center = np.mean(data, 1)

centers = np.matlib.repmat(center, N, 1)

data_cent = data - np.transpose(centers)

# Compute covariance and its eigenvalues from centered data

Sigma = np.cov(data_cent)

evals, evecs = np.linalg.eigh(Sigma)

# Return eigenvalues and eigenvectors

return np.flip(evals,0), np.flip(evecs, 1)

def __transformData(features, PCevecs):

return np.dot(features, PCevecs[:, 0:2])

PCevals, PCevecs = __PCA(trainingFeatures)

trainingFeatures2D = __transformData(trainingFeatures, PCevecs)

validationFeatures2D = __transformData(validationFeatures, PCevecs)

testingFeatures2D = __transformData(testingFeatures, PCevecs)

print('shape training = ', trainingFeatures2D.shape)

print('shape validation = ', validationFeatures2D.shape)

print('shape testing = ', testingFeatures2D.shape) # ----------------- Data (iris_new.csv):

6.012273103359637716e+00,2.921568019750800715e+00,5.144015868688889270e+00,1.970088604348821493e+00,2.000000000000000000e+00

6.206367020846028204e+00,3.228161981733847608e+00,5.003777773209909796e+00,2.070722282333296693e+00,2.000000000000000000e+00

5.625798030443882602e+00,3.904787165603675891e+00,1.357799778276990610e+00,4.936702532950165145e-01,0.000000000000000000e+00

6.808195662123572411e+00,3.178650674586387392e+00,5.100970180963188660e+00,1.889159208617503083e+00,1.000000000000000000e+00

6.581056019090129716e+00,2.913359774022858506e+00,4.587302757255009134e+00,1.544190829702803081e+00,1.000000000000000000e+00

4.922743626456581367e+00,3.535691449125819119e+00,2.132672532444197522e+00,4.239278473932364122e-01,0.000000000000000000e+00

5.802146320330803242e+00,2.946526640063509728e+00,4.959872418259350546e+00,2.021122281946927668e+00,2.000000000000000000e+00

6.375128323431418664e+00,2.666462047101149313e+00,5.821189071787672198e+00,1.536890095459532590e+00,2.000000000000000000e+00

7.533823868138515678e+00,3.132375225578859279e+00,6.572352764606229414e+00,1.905853718958894216e+00,2.000000000000000000e+00

6.829638868899861492e+00,3.331713498791738726e+00,4.813943469329544200e+00,1.586665861661761712e+00,1.000000000000000000e+00

5.950381506190458580e+00,2.832591299094304738e+00,5.124514747648298396e+00,2.075933042607615686e+00,2.000000000000000000e+00

6.698675459825690126e+00,3.013641660238998377e+00,5.871907459071210589e+00,2.316142751115663589e+00,2.000000000000000000e+00

5.404459128494982778e+00,3.457856717182044459e+00,1.539995605002054235e+00,3.332571255378059538e-01,0.000000000000000000e+00

4.495654321360945893e+00,3.305647532933800203e+00,1.584864478338738003e+00,3.858567310400071726e-01,0.000000000000000000e+00

5.339075543530506174e+00,3.973787538327554270e+00,1.634324363275604819e+00,5.430132463146704058e-01,0.000000000000000000e+00

7.601062852092417721e+00,3.002185731093820920e+00,6.343795144436067623e+00,2.139230252694762413e+00,2.000000000000000000e+00

6.351704787787541662e+00,3.429505864743360721e+00,6.014392461776881582e+00,2.566410057648921050e+00,2.000000000000000000e+00

6.379738060276616274e+00,2.632994102455576968e+00,5.275529757863798075e+00,2.047544577918903652e+00,2.000000000000000000e+00

5.741775363214967953e+00,4.513852190532458231e+00,1.690940065951078264e+00,5.135774649003405079e-01,0.000000000000000000e+00

5.505925961819941428e+00,3.122705091202445438e+00,4.543438336381168341e+00,1.664080033291209793e+00,1.000000000000000000e+00

7.801400975875588450e+00,3.168540458555272110e+00,6.116126874927260815e+00,2.426846823877313764e+00,2.000000000000000000e+00

5.357132049992936018e+00,3.983135840533841066e+00,1.662844363455381558e+00,3.625930788722653286e-01,0.000000000000000000e+00

7.019590355323099828e+00,3.122505046361530923e+00,5.029930044598924788e+00,1.547275107884078960e+00,1.000000000000000000e+00

6.911969636692916197e+00,3.392803025461516242e+00,6.155437090960832691e+00,2.541774303108266242e+00,2.000000000000000000e+00