What Is Gis

You might also like

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 3

WHAT IS GIS?

Geographic Information System (GIS) is a framework that is specifically used to collect, classify
as well as analyze data depending upon the geographical space time. The main commonality
among all the data points that are retrieved is that they ultimately refer to some physical
location on the earthly dimensions. With the advent of location services such as GPS, GIS has
become an integral part of almost every service that we render. By using GIS, organizations
around the world are now becoming masters in strategizing their operations as per the
demographics that they serve making the most out of the resources at hand. GIS relates
completely unrelated data to the geographical topography of earth, ultimately benefitting
almost every sector from agriculture to development of space technology.

USES OF MACHINE LEARNING IN GIS:


Machine learning have been used in building a variety of tools for geographical information
system analysis such as image classification tools, clustering tools etc. Some of them can be
listed as follows:-

 Image Classification:- It is one of the simplest implementation of machine learning with


GIS analysis systems. It can be used to classify geographical maps or geographical tagged
pictures according to the training data sets.
 Semantic Segmentation:- Semantic segmentation refers to the system of associating
each pixel of an image to a particular label. It is most commonly used for classifying
different land covers in geographical analysis systems.
 Instance Segmentation:- Instance segmentation can be defined as a super version of
object detection where the shape of an object is labeled and can be used later for
locating various places or structures of interests using satellite imagery. It has been
widely used by militaries and archaeological entities around the world.
 Object Detection:- Object detection refers to the process of labelling a particular object
and when combined with geographical information systems can be used to detect
objects using satellite imagery.

MACHINE LEARNING APPROACHES IN GIS:


There is an abundance of machine learning approaches that are being employed to provide
more accurate results in Geographical System Analysis. Some of them can be listed as follows:-

 Support Vector Machines:- These models are binary in nature and predict whether a
given set of input values belongs to either of the vectors given. Support vector machines
are mainly used for classification or regression but also for probabilistic functions in
conjunction with methods such as Platt scaling. It is most commonly used in the field of
satellite imagery classification. For example when we get the satellite imagery of a
particular location or demography, it is very much difficult to distinguish between
features just depending upon the shape and no elevation details. Both trees as well as
grasslands would appear to be green in nature and concrete structures such as buildings
or roads would appear as same. This is where support vector machines come into play.
Support vector machines are supervised machine learning approaches that are used
majorly for classification.
They basically mean creating a boundary on the hyperplane so as to divide data points
efficiently and in a manner, such that the distance between the hyperplane and the
nearest data point is maximized. Since they are unsupervised machine learning models,
we just need to train them with unlabeled data sets consisting of earthly features such
as trees, grasslands, concrete buildings, etc., and these models would eventually better
themselves with time in classifying between structures & bring about a revolution in
satellite imagery reading techniques.
 K means clustering:- K means clustering refers to one of the unsupervised machine
learning algorithms which takes an indigenous amount of unlabeled data as training
data sets and then classify them into clusters without the need of providing any
expected outcomes. K refers to the total number of target clusters that are required for
final segmentation and based upon this, the algorithm would divide the data sets into
different clusters with squared distances of each cluster. While dealing with satellite
imagery, k means clustering would cluster objects or features depending upon the
similarities between them. These similarities may refer to the featured characteristics of
the topography, number of similar structures such as concrete buildings and be further
enhanced to cluster locations based upon other factors such as number of schools in a
particular area or areas with high traffic or areas with high crime rate etc.
 Empirical Bayesian Networks:- Bayesian networks are models that provide a gist of an
event occurring and its potential outcome. They are acyclic in nature and represent a set
of random variables using a graph. They are most commonly used to identify the most
probable cause for the likelihood of an event to occur. Bayesian networks in
amalgamation with Kriging Interpolation, not only provide details about specific
activities or natural phenomena at a particular geo-location, but also make predictions
by building a mathematical function along with a feature called semi-variogram. These
Bayesian networks have proved to be really useful in correctly predicting various natural
phenomena such as rainfall in a particular place, heat waves around a particular
demography, etc., with existing training data sets to determine how they would vary in
the nearby feature and then cross validating the models.
WHAT HAS CHANGED IN THE RECENT YEARS?
There was a time when something such as cordless phones or touchscreen smartphones used
to startle us and were everyone’s admiration. But technology has advanced at a pace that
anyone could have ever imagined. The first industrial revolution took around 200 years, i.e.,
from the 1700s to the 1900s; the second revolution took about a hundred years, i.e., from the
early 1900s to the 2000s. The third revolution started with the new millennial and took hardly
ten to fifteen years to reach its peak. The technology advancements have certainly outpaced us
in the recent decades and what are the factors that have eased the process of integrating
machine learning processes with the geographical information systems. Some of such
advancing factors have been listed below:-

 Data Collection Abilities:- The world around us is becoming more and more data driven
day by day. There are now innumerable means to collect data and with the
advancement of technology and availability of sensory devices, drones and satellite
imagery, it has become much more easier to collect intrinsic data even from geospatial
perspective.
 Computing Abilities:-
 Algorithmic Improvements:-

You might also like