Building machine learning model

Hello,

I am having problem in creating machine learning model.

I try to use

fitcknn (), fitctree () 

to build a model.

However the fitcknn gives me some errors that I dont know how to fix.

Here is my code

load timefeat;
train = (time);
a = length(train);

timefeat_t=[];
for sg=1:a
    timefeat_t=[timefeat_t, train{1,sg}];
    Lengths_T(sg)=length(train{1,sg});
end
X=timefeat_t';

n_obs=size(timefeat_t,2);
y=cell(n_obs,1);
group1=sum(Lengths_T(1:1));
group2=group1+sum(Lengths_T(2:end));

y(1:group1)={'low speed'};                    % Class 1 definition - Train
y(group1+1:group2)={'high speed'};        % Class 2 definition - Train

X_Train=X;
Y_Train=y;
T_Train=table(X,y);

Model2_1=fitcknn(T_Train.X,T_Train.y,...
    'NumNeighbors',10,'Distance','cityblock');
Model2_2=fitctree(T_Train.X,T_Train.y);

I have included the data file "timefeat.mat"

Could anyone please check my code and help me understand what error did I make and how should i fix it?

Thank you

[keerthana pothula]

Machine Learning is a core component of Artificial Intelligence that includes how machines can analyze data, identify patterns and make decisions with low to no human intervention. With the ever-increasing demand for machine automated solutions ML has become one of the rapidly evolving technology along with AI & Data Science.

Find the Latest Machine Learning projects based on ML algorithms for open source machine learning.

Facial Emotion Detection using Neural Networks

Cancer Prediction using Naive Bayes

Tour Recommender App Using Collaborative Filtering

Driver Drowsiness Detection System for Accident Prevention

Automatic Salt Segmentation with UNET in Python using

Deep Learning Transformer Conversational Chatbot in Python using TensorFlow

2.0 Lane-Line Detection System in Python using OpenCV

Facial Emotion Recognition and Detection in Python using Deep Learning

Diabetes Prediction Using Data Mining

Data Mining for Sales Prediction in Tourism Industry

Higher Education Access Prediction

Software Hotel Recommendation System Based on Hybrid Recommendation Model

Answer 1

[Nipun Katyal]

Hi, as you have only two labels such a large number of neighbours are not required, Instead you can change the distance to 'cosine' and try shuffling the rows. You can find improved result using the configurations below,

Model2_1=fitcknn(T_Train.X,T_Train.y,...
    'NumNeighbors',5,'Distance','cosine');

[An Van]

Thank you, after changing the define, it works