Deep insight of K-Nearest Neighbour(KNN) | The Lazy algorithm | Mathematics | Machine Learning

-

Deep insight of K-Nearest Neighbour(KNN) | The Lazy algorithm | Mathematics | Machine Learning

Photo by Nina Strehl on Unsplash


It is an instance-based supervised machine learning algorithm. In this algorithm, we find what is the neighbour instance of this value. It is used for both regression and classification. 

KNN Classification:

Let's have a look at the following diagram, and focus on the Heart-shaped Marker(💜), try to examine visually that in which category it will lie in red class or blue class.

The Graph

If we are assuming 5 nearest neighbours around the Heart, then visually you can determine that is more likely to belong to the blue class. Because there are almost three points are the nearest neighbour of the heart from the blue class and only two points from the red class. See the Graph.

The Graph


So, this is all about the K-nearest Neighbour classification.
K-nearest neighbour is a simple supervised machine learning algorithm, that stores the all available cases and classifies new cases and classifies new cases on a similarity measure(e.g., distance function).

Step by Step Algo:

1. Choose the "K" values, that is how many nearest neighbours we have to consider. This step should be done carefully because the number of nearest value i.e. you are considering is the quality factor for your model. It directly affects your model's training.

2. We calculate the distance of the neighbours around the given point and then select the K nearest neighbours only. With the help of the distance formula:

Distance Formula


3. How many neighbours are related to which category? The majority will be selected as a result. And by only just three steps we can classify that the given point belongs to which category.

Finding K value: 

  • For finding the K value we can use hit and trial method, just put random values again and again and then check your model accuracy. But it will not work for large datasets.
  • Always go for a larger value of K, because it will decrease the noise in results, and in the end it will give you better accuracy.
  • Last but not least, you can do cross-validation. Just take a small part of your dataset and use it to find the best value of K and then apply it to the whole dataset.

If you love my work then you can connect me on LinkedIn and Github.

Comments

Post a Comment

Popular posts from this blog

SMART HOSPITAL | IoT based smart health monitoring system | Real time responses | Internet Of Things | ESP8266

-
Image
SMART HOSPITAL | IoT based smart health monitoring system | Real time responses | Internet Of Things | ESP8266 Problem Statement                                    Our problem statement is now a day many busy people are facing more problems regarding their health due to their busy lives they are not to able to monitor their health condition properly. And they get easily sick or committed to weakness just because they are not able  to update their health and sometimes if we did not mind them then these small diseases can be fatal in the future. So we have to make a system which can help them out to monitor their own health problems and all the received data can be used in the treatment of that individual. Introduction According to our theme i.e. Smart Hospital , we will make our health c...
Read more

A Quick Guide to Data pre-processing for Machine Learning | Python | IMPUTATION | STANDARDISATION | Data Analysis | Data Science

-
Image
Data Pre-processing | Imputation | Standardisation | Rescaling | Python Before feeding your data to the machines you have to prepare this data before inserting it into your Machine Learning Algorithm. Quality of your data has to be good, it doesn't contain null values or out of the constraint values. Because the quality of your data is directly related to the quality of training of your model. For more info about Data-preprocessing's need and requirements,  Click Here. Photo by  Mika Baumeister  on  Unsplash 1. Imputation Imputation's simply meant the "change", this process will help you to change the missing value from your table. There are lots of algorithms which can't deal with the null values and might give you errors or badly trained model. Let's have a look at the data below, this data is about the salary of several domains... THE DATASET This data has some null values, which is unbearable by the machine learning ...
Read more

Plotly & Cufflinks | A Data Visualisation Library with Modern Features | Python | Data Science | Data Visualisation

-
Image
Plotly & Cufflinks | A Data Visualisation Library with Modern Features | Python | Data Science | Data Visualisation Plotly Plotly gives you lots of interactive and dynamic data visualisation & UI tools for data science, Machine Learning & Engineering. It has some great features which we will discuss in this post. So, be ready for the tutorial. Installation Guide on your Machine: For Plotly put this on your terminal and run. pip install plotly or conda install -c plotly plotly Cufflinks For using Plotly library we have to configure it with the pandas, so the cufflinks library giver us a privilege to do it. Because the Plotly is built on the top of d3.js . Installation Guide on your Machine: For cufflinks put this on your terminal and run. pip install cufflinks or conda install -c conda-forge cufflinks-py Let's begin... Import all the Libraries needed. import pandas as pd import numpy as np import cufflinks as cf from plotly.offline import download_plotlyjs,init_notebook_...
Read more