UBER RELATED DATA ANALYSIS USING

MACHINE LEARNING

Submitted in partial fulfilment of the requirements for the award of

Bachelor of Engineering Degree inComputer Science and Engineering

By
RISHI.S
(Reg.No. 37110642)

DEPARTMENT OF COMPUTER SCIENCE AND ENGINEERING

SCHOOL OF COMPUTING

SATHYABAMA
INSTITUTE OF SCIENCE AND TECHNOLOGY
(DEEMED TO BE UNIVERSITY)
Accredited with Grade “A” by NAAC
JEPPIAAR NAGAR, RAJIV GANDHI SALAI,
CHENNAI-6000119, TAMIL NADU

APRIL 2021
 SATHYABAMA
INSTITUTE OF SCIENCE AND TECHNOLOGY
(DEEMED TO BE UNIVERSITY)
Accredited with ―A‖ grade by NAAC
Jeppiaar Nagar, Rajiv Gandhi Salai, Chennai - 600119
www.sathyabama.ac.in

DEPARTMENT OF COMPUTER SCIENCE ENGINEERING

BONAFIDE CERTIFICATE

This is to certify that this Professional Training Report is the bonafide work of RISHI.S
(37110642) who underwent the professional training in―Uber Related Data Analysis
using Machine Learning‖ underour supervision from November 2020 to April2021.

Internal Guide

DR.B.ANKAYARKANNI,M.E, Ph.D

Head of the Department

Submitted for Viva Voce Examination held on

InternalExaminer ExternalExaminer
 DECLARATION

I, RISHI.S (37110642), hereby declare that the Professional Training Reporton ―UBER
RELATED DATA ANALYSIS USING MACHINE LEARNING”,done by me under the
guidance of DR.B.ANKAYARKANNI ,M.E,Ph.D atSathyabama Institute of Science and
Technology, is submitted in partial fulfilment of the requirements for the award of
Bachelor of Engineering degree in Computer Science.

DATE:
PLACE:CHENNAI SIGNATURE OF THE CANDIDATE
 ACKNOWLEDGEMENT

I am pleased to acknowledge my sincere thanks to Board of Management of

SATHYABAMA for their kind encouragement in doing this project and for
completing it successfully. I am grateful to them.

I convey my thanks to Dr. T.Sasikala M.E.,Ph.D., Dean, School of Computing

Dr.S.Vigneshwari M.E., Ph.D. and Dr.L.Lakshmanan M.E., Ph.D. , Heads of the
Department of Computer Science and Engineering for providing us necessary
support and details at the right time during the progressive reviews.
I would like to express my sincere and a deep sense of gratitude to my Project
Guide Dr. B. Ankayarkanni M.E., Ph.D., for her valuable guidance, suggestions
and constant encouragement paved way for the successful completion of my
project work.

I wish to express my thanks to all Teaching and Non-teaching staff members of the
Departmentof Computer Science and Engineering who were helpful in many
ways for the completion of the project.
 ABSTRACT

The project ―Uber Related Data Analysis using Machine Learning‖ explains the
working of an Uber dataset, which contains data produced by Uber for New York
City. Uber is defined as a P2P platform. The platform links you to drivers who can
take you to your destination. The dataset includes primary data on Uber pickups
with details including the date, time of the ride as well as longitude-latitude
information, Using the information, the paper explains the use of the k-means
clustering algorithm on the set of data and classify the various parts of New York
City. Since the industry is booming and expected to grow shortly. Effective taxi
dispatching will facilitate each driver and passenger to reduce the wait time to
seek out one another. The model is employed to predict the demand on points of
the city.
 TABLE OF CONTENTS

CHAPTER No. TITLE PAGE No.

ABSTRACT 7

LIST OF ABBREVIATIONS 10
LIST OF FIGURES 11

1 INTRODUCTION 12
1.1 Introductionto Uber Dataand 12
Predictions
1.1.1 Machine Learning Predictions 12
Introduction to Machine Learning13
Overview 13
Machine Learning Approaches 14
Theory 14
Models 15
2 LITERATURE REVIEW 18

2.1 Related Works 18

3 AIM AND SCOPE OF PROJECT 21

Aim Of Project 21
Objectives 21
Scope Of Project 21
4 MODULE IMPLEMENTATIONS 22
Modules 22
%Pylab Inline 22
Pandas 22
Seaborn 23
Kmeans 24
YellowBrick 24
Folium 24
Modules Hardware/Software 24
 Architectural Design 25
Module Implementations 26
5 RESULTS AND DISCUSSION 27
6 CONCLUSION AND FUTURE WORK 33
REFERENCES
APPENDIX

A. PLAGAIRISM REPORT
B.BASE PAPER
C. SOURCE CODE
 LIST OF ABBREVIATIONS

ABBREVIATIONS EXPANSION
ML Machine Learning

ANN Artificial Neural Network

GA Genetic Algorithms

DOM Date of Month

 LIST OF FIGURES

FIGURE No. FIGURE NAME PAGE No.

4.1 ARCHITECTURAL DIAGRAM 24

IMPORTING MODULES 27
IMPORING DATASET 27
FINAL DATASET 28
VISUALISATION OF X-Y GRAPH FREQUENCY OF CABS 28
VISUALIZATION OF DATA BASED ON MONTH 29
VISUALIZATION OF DATA BASED ON DAY 29
PLOTTING HEAT MAP OF DATASET 29
CROSS REFERENCING LATITUDE AND LONGITUDE 30
PLOTTING COLLECTION OF POINTS IN XY GRAPH 30
GETTING THE DATA OF CENTROIDS FROM ALGORITHM 30
PLOTTING THE CENTROIDS ON MAP 31
TOTAL NUMBER OF TRIPS BASED ON CLUSTERS 31

PREDICTION OF PICKUP OF CAB BASED ON CLUSTER 32

 CHAPTER 1
INTRODUCTION

INTRODUCTION TO UBER DATA AND PREDICTIONS

Uber Technologies, Inc., commonly called Uber, is an American technology

company. Its services include ride-hailing, food delivery (Uber Eats), package
delivery, couriers, freight transportation, and, through a partnership with Lime,
electric bicycle and motorized scooter rental. Uber Technologies may be a P2P
network for sharing travel The Uber platform connects you with drivers who can
take you to your destination or location. This dataset includes primary data on
Uber collections with details that include the date, time of travel, yet as information
on longitude and latitude in urban center and has operations in over 900
metropolitan areas worldwide. The prediction of the frequency of trips of
knowledge is by implementing an element of k-means clustering algorithm,the
quality algorithm describes the most variance within the group because the
number of square distances Euclidean distances between the points and also the
corresponding centroid.

Machine Learning Predictions

With the appearance of the computing device, uber data prediction has since
moved into the technological realm. the foremost prominent technique involves the
employment of artificial neural networks (ANNs) and Genetic Algorithms(GA).
Scholars found bacterial chemotaxis optimization method may perform better than
GA. ANNs is thought of as function approximators. the foremost common sort of
ANN in use for stock exchange prediction is that the feed forward network utilizing
the backward propagation of errors algorithm to update the network weights.
These networks are commonly cited as Backpropagation networks. Another type
of ANN that's more appropriate for uber data prediction is that the time recurrent
neural network (RNN) or time delay neural network (TDNN). samples of RNN and
TDNN For uber data prediction with ANNs, there are usually two approaches
taken for forecasting different time horizons: independent and joint. The
independent approach employs one ANN for every time horizon, for instance, 1-
day, 2-day, or 5-day. The advantage of this approach is that network forecasting
error for one horizon won't impact the error for one more horizon—since every
time horizon is usually a novel problem. The joint approach, however, incorporates
multiple time horizons together so they're determined simultaneously. during this
approach, forecasting error for just once horizon may share its error therewith of
another horizon, which might decrease performance. There also are more
parameters required for a joint model, which increases the chance of overfitting.
INTRODUCTION TO MACHINE LEARNING
Machine learning (ML) is that the study of computer algorithms that improve
automatically through experience. it's seen as a subset of computing. Machine
learning algorithms build a mathematical model to support sample data, called
"training data", so on form predictions or decisions without being explicitly
programmed to undertake to do so. Machine learning algorithms are utilized in a
very large sort of applications, like email filtering and computer vision, where it's
difficult or infeasible to develop conventional algorithms to perform the needed
tasks.

Overview
Machine learning involves computers discovering how they'll perform tasks without
being explicitly programmed to try to to so. It involves computers learning from
data provided so they perform certain tasks. for easy tasks assigned to computers,
it's possible to program algorithms telling the machine the way to execute all steps
required to unravel the matter at hand; on the computer's part, no learning is
required. For more advanced tasks, it will be challenging for an individual's to
manually create the needed algorithms. In practice, it can end up to be simpler to
assist the machine develop its own algorithm, instead of having human
programmers specify every needed step.

The discipline of machine learning employs various approaches to show

computers to accomplish tasks where no fully satisfactory algorithm is on the
market. In cases where vast numbers of potential answers exist, one approach is
to label a number of the proper answers as valid. this may then be used as
training data for the pc to boost the algorithm(s) it uses to work out correct
answers. as an example, to coach a system for the task of digital character
recognition, the MNIST dataset of handwritten digits has often been used
 Machine Learning Approaches

Machine learning approaches are traditionally divided into three broad categories,
counting on the character of the "signal" or "feedback" available to the training
system:

• Supervised learning: the pc is presented with example inputs and their desired
outputs, given by a "teacher", and so the goal is to be told a general rule that maps
inputs to outputs.

• Unsupervised learning: No labels are given to the training algorithm, leaving it on

its own to go looking out structure in its input. Unsupervised learning could be a
goal in itself (discovering hidden patterns in data) or how towards an end (feature
learning).

• Reinforcement learning: A worm interacts with a dynamic environment during

which it must perform a specific goal (such as driving a vehicle or playing a game
against an opponent). because it navigates its problem space, the program is
provided feedback that's analogous to rewards, which it tries to maximise.

Other approaches are developed which don't fit neatly into this three-fold
categorisation, and sometimes over one is used by the identical machine learning
system. as an example topic modelling, dimensionality reduction or meta learning.

As of 2020, deep learning has become the dominant approach for ongoing arenas
of machine learning.

Theory

A core objective of a learner is to generalize from its experience. Generalization

during this context is that the ability of a learning machine to perform accurately on
new, unseen examples/tasks after having experienced a learning data set. The
training examples come from some generally unknown probability distribution
(considered representative of the space of occurrences) and therefore the learner
has got to build a general model about this space that permits it to supply
sufficiently accurate predictions in new cases.
The computational analysis of machine learning algorithms and their performance
could be a branch of theoretical engineering science referred to as computational
learning theory. Because training sets are finite and therefore the future is
uncertain, learning theory usually doesn't yield guarantees of the performance of
algorithms. Instead, probabilistic bounds on the performance are quite common.
The bias–variance decomposition is a method to quantify generalization error.
For the simplest performance within the context of generalization, the complexity
of the hypothesis should match the complexity of the function underlying the
information. If the hypothesis is a smaller amount complex than the function, then
the model has under fitted the information. If the complexity of the model is
increased in response, then the training error decreases. But if the hypothesis is
simply too complex, then the model is subject to overfitting and generalization are
going to be poorer.
In addition to performance bounds, learning theorists study the time complexity
and feasibility of learning. In computational learning theory, a computation is taken
into account feasible if it is tired polynomial time. There are two types of time
complexity results. Positive results show that a specific class of functions is
learned in polynomial time. Negative results show that certain classes can't be
learned in polynomial time.
Models
ARTIFICIAL NEURAL NETWORKS
Artificial neural networks (ANNs), or connectionist systems, are computing
systems vaguely inspired by the biological neural networks that constitute animal
brains. Such systems "learn" to perform tasks by considering examples, generally
without being programmed with any task-specific rules.

An ANN may be a model supported a group of connected units or nodes called

"artificial neurons", which loosely model the neurons in a very biological brain.
Each connection, just like the synapses during a biological brain, can transmit
information, a "signal", from one artificial neuron to a different. a man-made
neuron that receives a sign can process it and so signal additional artificial
neurons connected to that. In common ANN implementations, the signal at a
connection between artificial neurons could be a imaginary number, and therefore
the output of every artificial neuron is computed by some non-linear function of the
sum of its inputs. The connections between artificial neurons are called "edges".
Artificial neurons and edges typically have a weight that adjusts as learning
proceeds. the burden increases or decreases the strength of the signal at a
connection. Artificial neurons may have a threshold specified the signal is barely
sent if the mixture signal crosses that threshold. Typically, artificial neurons are
aggregated into layers. Different layers may perform different sorts of
transformations on their inputs. Signals travel from the primary layer (the input
layer) to the last layer (the output layer), possibly after traversing the layers
multiple times.
The original goal of the ANN approach was to resolve problems within the same
way that somebody's brain would. However, over time, attention moved to
performing specific tasks, resulting in deviations from biology. Artificial neural
networks are used on a spread of tasks, including computer vision, speech
recognition, MT, social network filtering, playing board and video games and
diagnosing.

Deep learning consists of multiple hidden layers in a synthetic neural network. This
approach tries to model the way the human brain processes light and sound into
vision and hearing. Some successful applications of deep learning are computer
vision and speech recognition.

REGRESSION ANALYSIS

Regression analysis encompasses an outsized style of statistical methods to

estimate the link between input variables and their associated features.
Its commonest form is statistical regression, where one line is drawn to best fit the
given data in step with a mathematical criterion like ordinary statistical procedure.
The latter is usually extended by regularization (mathematics) methods to mitigate
overfitting and bias, as in ridge regression. When managing non-linear problems,
go-to models include polynomial regression (for example, used for trendline fitting
in Microsoft Excel), Logistic regression (often employed in statistical
classification) or maybe kernel regression, which introduces non-linearity by taking
advantage of the kernel trick to implicitly map input variables to higher dimensional
space.
K-MEANS CLUSTERING

k-means clustering could be a method of vector quantization, originally from signal

processing, that aims to partition n observations into k clusters within which each
observation belongs to the cluster with the closest mean (cluster centres or cluster
centroid), serving as a prototype of the cluster. This leads to a partitioning of the
info space into Voronoi cells. k-means clustering minimizes within-cluster
variances (squared Euclidean distances), but not regular Euclidean distances,
which might be the harder Weber problem: the mean optimizes squared errors,
whereas only the geometric median minimizes Euclidean distances. for example,
better Euclidean solutions may be found using k-medians and k-medoids.

The problem is computationally difficult (NP-hard); however, efficient heuristic

algorithms converge quickly to an area optimum. These are usually kind of like the
expectation-maximization algorithm for mixtures of Gaussian distributions via an
iterative refinement approach employed by both k-means and Gaussian mixture
modelling. They both use cluster centers to model the data; however, k-means
clustering tends to search out clusters of comparable spatial extent, while the
expectation-maximization mechanism allows clusters to own different shapes.

The unsupervised k-means algorithm includes a loose relationship to the k-nearest

neighbour classifier, a preferred supervised machine learning technique for
classification that's often confused with k-means thanks to the name. Applying the
1-nearest neighbor classifier to the cluster centers obtained by k-means classifies
new data into the present clusters. this can be referred to as nearest centroid
classifier or Rocchio algorithm.
 CHAPTER 2

LITERATURE REVIEW

2.1 RELATED WORKS

A state in which the results, k-means clustering is used to estimate the most likely
collection points at a given time and to predict the best hotspots of nightlife
learning trends from previous Uber pickups. This has been verified using the Lyft
test set and is consistent with Yelp's best results.[1]

Poulsen, L.K In this document, they conducted a spatial analysis of the Green Cab
and Uber races in the outer districts of New York City to determine the competitive
position of the NYCTLC. We found that the demand for green taxis continues to
increase, but the number of Uber trips in the same area is growing faster.
However, the analysis showed that in Greece, in an area that is generally poor in
general. They did not find any variation between the green taxi and Uber when
variations were observed between weekdays and weekends. This research
recommends that NYCTLC create a dashboard that analyzes and displays data in
real time, as we believe this will increase its competitiveness compared to Uber.
Uber is a recent taxi operator in New York and is constantly devouring the market
share of the yellow and green taxis of the New York Taxi and Limousine
Commission (NYCTLC)[2]

Faghih, S.S recommends a recent modeling approach in Manhattan, New York

City, to capture the demand for electronic mail services, particularly the Uber
application. Uber collection data is added to the Manhattan TAD level and at 15-
minute time intervals. This aggregation allows the implementation of a modern
approach to spatio-temporal modeling to obtain a spatial and temporal
understanding of the demand. During a typical day, two spacetime models were
developed using Uber collection data, the STAR and STAR and MSPE turns
determine the output of the models. The results of the MSPE have shown that it is
recommended to use the Lasso-Star system instead of the star design. A
comparison between the demand for yellow and uber taxis in 2014 and 2015 in
New York shows that the demand for uber has increased.[3]
Ghuha declared the grouping of the sequence observed using a small amount of
memory and time. The data flow model is necessary for new kinds of applications
that involve large data sets, such as web click flow analysis and multimedia data
flow data model analysis where: data A sequence of points, the objective is
Maintain a constant level. In the document, we will consider models in which the
groups have a different point or "center". In the k-median question, the objective is
to minimize the average distance from the data points to the nearest cluster
centers.[4]

Ahmed, M., has shown that by using detailed data on taxis at the travel level and
on the rental vehicle and data on complaints about the level of new complaints at
the level of incidents, we study how Uber and Lyft enter damaged the quality of
taxi services in New York City. The overall effect of the organizations based on the
scenario and in particular of the riding administrations was enormous and
widespread. One of these effects is the expansion of the rivalry between Uber and
Lyft over the quality of taxi administration. They use a new set of complaint data to
measure (the lack of) quality of service that we have never been analyzed before.
Focus on the quality dimensions generated by most of the complaints we
demonstrate. The increased competition for these shared travel services has had
an intuitive impact on the behavior of taxi drivers[5]

Wallsten, S, stated that the results of New York and Chicago are consistent with
the possibility that taxis react to the new challenge by improving quality. In New
York, the rise of Uber is linked to the reduction of objections to travel to the city.
They discuss the competitive effect of sharing taxis in the taxi industry using the
complete data set of the New York City Taxi and Limousine Commission for more
than one billion taxi trips in complaints and details of New York, New York and
Chicago Google Trends on the success of Uber's largest shared travel service.[6]

Sotiropoulos, D.N, represented that this document addresses the problem of

grouping, by using a new approach to genetic algorithms that is highly scalable in
large volumes of textual details, developing a coding scheme based on centroids.
We apply k means clustering algorithm in this document. Clustering is the
unsupervised machine learning algorithm used to solve grouping problems based
on similarities. This technique has aroused interest in a wide range of scientific
fields, which address clustering methods, to solve complex classification
problems.[7]
 CHAPTER 3
AIM AND SCOPE OF PROJECT

AIM OF PROJECT
The aim of the project is to predict the pickup of the cab with respect to the
location given by the user in the app from clusters which are main coordinated
points in a city by applying data visualization on the basis of frequency of trips
travelled with respect to the day in the month and applying data analysis using the
algorithm of k-means clustering.

OBJECTIVES

The objective of the project is to plot a heat map for the prediction of frequency of
trips across a collection of points(latitude,longitude)in a dataset which is data
visualization and using these objectives apply the data in the algorithm adopted in
the project.

SCOPE OF PROJECT

Based on the aim and the objectives in the project, the scope of the project is to
apply an algorithm best suited to analyze the data imported from the dataset such
that the algorithm can train the data and then proceed for data analysis.

.
 CHAPTER 4

MODULE IMPLEMENTATION

MODULES

The modules used are based on the process of data acquisition,processing and
analysis of a dataset.
The dataset includes primary data on Uber pick-ups with details including the date,
time of the ride as well as longitude-latitude information of the city.
The dataset is imported and then analyzed and computized by analysis by
clustering algorithms,(K-means clustering). The module description can be
explained by importing the following modules from python.

%PYLAB INLINE
PyLab can be defined as a procedural interface to the Matplotlib object-oriented
plotting libraryMatplotlib which is that the whole package; matplotlib.pyplot can be
defined as a module in Matplotlib; and PyLab can be defined as a module that
gets installed alongside Matplotlib.
PyLab can be defined as a convenience module that bulk imports matplotlib.pyplot
(for plotting) and NumPy (for Mathematics and dealing with arrays) in a very single
name space.

PANDAS
In computer programing, pandas can be defined as a software library written for
the Python programing language for data manipulation and analysis. particularly, it
offers data structures and operations for manipulating numerical tables and
statistics and it's free software was released under the three-clause BSD license.
Pandas is principally used for data analysis. Pandas allows importing data from
various file formats like comma-separated values, JSON, SQL, Microsoft Excel.
Pandas allows various data manipulation operations like merging, reshaping,
selecting, still as data cleaning, and data wrangling features.
 SEABORN

Seaborn is a library for making statistical graphics in Python. It builds on top of

matplotlib and integrates closely with pandas data structures.

Seaborn helps you explore and understand your data. Its plotting functions
operate on dataframes and arrays containing whole datasets and internally
perform the necessary semantic mapping and statistical aggregation to produce
informative plots. Its dataset-oriented, declarative API lets you focus on what the
different elements of your plots mean, rather than on the details of how to draw
them.
It gives us the capability to create amplified data visuals. This helps us understand
the data by displaying it in a visual context to unearth any hidden correlations
between variables or trends that might not be obvious initially. Seaborn has a high-
level interface as compared to the low level of Matplotlib.
Seaborn comes with a large number of high-level interfaces and customized
themes that matplotlib lacks as it’s not easy to figure out the settings that make
plots attractive. Matplotlib functions don’t work well with dataframes, whereas
seaborn does.
Seaborn’s integration with matplotlib allows you to use it across the many
environments that matplotlib supports, inlcuding exploratory analysis in notebooks,
real-time interaction in GUI applications, and archival output in a number of raster
and vector formats.

While you can be productive using only seaborn functions, full customization of
your graphics will require some knowledge of matplotlib’s concepts and API. One
aspect of the learning curve for new users of seaborn will be knowing when
dropping down to the matplotlib layer is necessary to achieve a particular
customization. On the other hand, users coming from matplotlib will find that much
of their knowledge transfers.
 KMEANS

The k means module is an unsupervised machine learning technique module

accustomed to identify clusters of information objects in an exceeding dataset.

The module is employed effectively and goal of the algorithm is to seek out groups
within the data, with the quantity of groups represented by the varaiable K. Data
points are clustered which is supported by the feature similarity.
YELLOWBRICK

Yellowbrick can be defined as a suite of visual analysis and diagnostic tools

designed to facilitate machine learning with scikit-learn. The library implements a
replacement core API object, the Visualizer which is that's an scikit-learn estimator
— an object that learns from data. almost like transformers or models, visualizers
learn from data by creating a visible representation of the model selection
workflow.

The Visualizer allow users to steer the model selection process, building intuitions
around feature engineering, algorithm selections and hyperparameter tuning. for
example, they'll help diagnose common problems surrounding model complexity
and bias, heteroscedasticity, underfit and overtraining, or class balance issues. By
applying visualizers to the model selection workflow, It allows you to steer
predictive models toward more successful results, faster.

FOLIUM

Folium can be defined as a module and a tool for plotting maps, it's a strong
library that helps you create several sorts of Leaflet maps.The Folium results are
interactive as it makes the library very useful for dashboard building. By
default,Folium creates a map in an exceedingly separate HTML file and plots the
info obtained from the dataset

MODULE SOFTWARE/HARDWARE
Software Configuration:
 Anaconda Navigator
 Jupyter Notebook
 Google Chrome
 Hardware Configuration:
 Windows 10,8,7
 Windows Server 2019 or Windows Server 2016
 Memory of atleast 4GB RAM
 Storage of Atleast 128 GB
 1 GB Graphics Card

ARCHITECTURAL DESIGN

Fig 4.1: ARCHITECTURE DIAGRAM

MODULE IMPLEMENTATIONS

 The module is based on data acquisition, processing and analysis of a

dataset.

 The dataset includes primary data on Uber pick-ups with details including
the date, time of the ride as well as longitude-latitude information of the city
.
 The dataset is imported and then analyzed and computized by analysis by
clustering algorithms,(K-means clustering).

 Import the packages numpy and seaborn and import the dataset from
Kaggle or github(uber.csv) which contains the dataset

 Import the data in jupyter notebook and using the data in dataset we
calculate the frequency of trips in an given area in this dataset (New York
City) from the latitudes and longitudes in the dataset

 Generate the heatmap and predict the pinpoint locations of travel of the cab

 Import the required packages required for analysis i.e sklearn,yellowbrick

and folium

 Categorize the centroids on the basis of latitude and longitude

 Mark the centroid points on the map imported from the module folium

 Predict the cluster from where the cab is scheduled to pickup from the data.
 CHAPTER 5
RESULTS AND DISCUSSIONS

As a result of this system getting introduced in our data analysis, it will act like a
mechanism to assign a cab to a customer to cover the gap of communication
between the company and the customer. The company will be able to understand
the customer better by deploying more cabs based on the hotspots plotted on the
map using the algorithm.

FIG 5.1 IMPORTING MODULES

Figure 5.1 explains the importing of modules in the notebook after importing the
modules we can import the dataset

FIG 5.2 IMPORTING DATASET

After importing the dataset,add more columns such as dom,weekday and hour
 FIG 5.3 FINAL DATASET
After importing 5.3,let us visualize the data on the basis of frequency of cabs
travelled in a day with respect to day,month.

FIG 5.4 VISUALISATION OF X-Y GRAPH OF FREQUENCY OF CABS

 FIG 5.5 VISUALIZATION OF DATA BASED ON MONTH

FIG 5.6 VISUALIZATION OF DATA BASED ON DAY

FIG 5.7 PLOTTING HEAT MAP OF DATASET

 FIG 5.8 CROSS REFERENCING LATITUDE AND LONGITUDE

FIG 5.9 PLOTTING COLLECTION OF POINTS IN XY GRAPH

FIG 5.10 GETTING THE DATA OF CENTROIDS FROM ALGORITHM

 FIG 5.11 PLOTTING THE CENTROIDS ON MAP

Based on figures 5.4 to 5.11 we are discussing how the data is visualized on the
basis of day,month and plotting all the points of the dataset on a x-y graph
referencing latitude and longitude after plotting the points we proceed for data
analysis using the algorithm.

Figure 5.10 shows taking the data from the dataset and using the algorithm to
calculate the centroids using k-means algorithm and plotting the points on the
map.

FIG 5.12TOTAL NUMBER OF TRIPS BASED ON CLUSTERS

Based on the data collected by the means of klabels by labelling the data in the
dataset in the form of clusters. Fig 5.11 explains the total number of trips on the
basis of clusters.
 FIG 5.13 PREDICTION OF PICKUP OF CAB BASED ON THE CLUSTER

Based on the figures 5.1 to 5.12 we can discuss that using data analysis the
program can predict the pickup location of the cab based on clusters analyzed and
applied by k-means clustering to schedule the cab for pickup.
 CHAPTER 5
CONCLUSION AND FUTURE WORK

CONCLUSION
This program will make the system of deploying more cabs to the required location
and makes it flexible for users. The users have no need to worry about the location
as the program will help in scheduling a cab for pickup nearest to the location. The
program shows the concepts of machine learning such as data visualization and
data analysis which makes the program and efficient for future work

FUTURE WORK
In future, system will provide the location of pickup to the users. Users can send
their location to the app, and the program used in the project will predict the
nearest location to the user and assign a cab to the user. The program and data
elements in the program developed must be tested by Uber such that it can be
used as an operational environment. It will make the program of predicting the
trips using data analysis more flexible and efficient for users.
.
 REFERENCES

[1]. Poulsen, L.K., Dekkers, D., Wagenaar, N., Snijders, W., Lewinsky, B.,
Mukkamala, R.R. and Vatrapu, R., 2016, June. Green Cabs vs. Uber in New York
City. In 2016 IEEE International Congress on Big Data (BigData Congress) (pp.
222-229). IEEE.

[2]. Verma, N. and Baliyan, N., 2017, July. PAM clusteringbased taxi hotspot
detection for informed driving. In 2017 8th International Conference on Computing,
Communication and Networking Technologies (ICCCNT) (pp. 1-7). IEEE.

[3]. Sotiropoulos, D.N., Pournarakis, D.E. and Giaglis, G.M., 2016, July. A genetic
algorithm approach for topic clustering: A centroid-based encoding scheme. In
2016 7th International Conference on Information, Intelligence, Systems &
Applications (IISA) (pp. 1-8). IEEE.

[4]. Guha, S. and Mishra, N., 2016. Clustering data streams. In Data stream
management (pp.169-187). Springer, Berlin, Heidelberg.

[5]. Shah, D., Kumaran, A., Sen, R. and Kumaraguru, P., 2019, May. Travel Time
EstimationAccuracy in Developing Regions: An Empirical Case Study with Uber
Data in Delhi-NCR✱.In Companion Proceedings of The 2019 World Wide Web
Conference (pp. 130-136). ACM.

[6]. Ahmed, M., Johnson, E.B. and Kim, B.C., 2018. The Impact of Uber and Lyft
on TaxiService Quality Evidence from New York City. Available at SSRN 3267082.

[7]. Wallsten, S., 2015. The competitive effects of the sharing economy: how is
Uber changing taxis. Technology Policy Institute, 22, pp.1-21.
[8]. Verma, N. and Baliyan, N., 2017, July. PAM clustering based taxi hotspot
detection forinformed driving. In 2017 8th International Conference on Computing,
Communication andNetworking Technologies (ICCCNT) (pp. 1-7). IEEE.

[9]. Kumar, A., Surana, J., Kapoor, M. and Nahar, P.A., CSE 255 Assignment II
Perfecting Passenger Pickups: An Uber Case Study.
 APPENDIX

A) PLAGIARISM REPORT

UberrelatedDataAnalysis (1).pdf
ORIGINALITY REPORT

8 %
SIMILARITY INDEX
6%
INTERNET SOURCES
4%
PUBLICATIONS
3%
STUDENT PAPERS

PRIMARY SOURCES

1
www.simplilearn.com
Internet Source
3%
2
cseweb.ucsd.edu
Internet Source
2%
G. Anitha, Mohamed Ismail, S.K.
3
Lakshmanaprabu. "Identification and 1%
characterisation of choroidal neovascularisation
using e-Health data through an optimal
classifier", Electronic Government, an
International Journal, 2020
Publication

Lasse Korsholm Poulsen, Daan Dekkers,

4
Nicolaas Wagenaar, Wesley Snijders, Ben 1%
Lewinsky, Raghava Rao Mukkamala, Ravi
Vatrapu. "Green Cabs vs. Uber in New York
City", 2016 IEEE International Congress on Big
Data (BigData Congress), 2016
Publication

5
www.ubdc.ac.uk
Internet Source
1%
B) BASE PAPER

Uber Related Data Analysis using Machine Learning

Rishi Srinivas 1 B. Ankayarkanni2
2
1
UG Student, Department of CSE, Sathyabama Institute of Associate Professor,Department of CSE, Sathyabama
Science and Technology,Chennai.India. Institute of Science and Technology,India.
2
1
[email protected], [email protected],

Abstract-The paper explains the working of an Uber The ultimate aim of the project is to predict
dataset, which contains data produced by Uber for New the pickup of the cab on the basis of clusters defined
York City. Uber is defined as a P2P platform. The by the k-means clustering algorithm. This algorithm
platform links you to drivers who can take you to your is used to divide the dataset into k-groups. where k is
destination. The dataset includes primary data on Uber defined as the number of groups provided by the
pickups with details including the date, time of the ride user. The standard algorithm describes the maximum
as well as longitude-latitude information , Using the variance within the group as the number of square
information, the paper explains the use of the k-means
clustering algorithm on the set of data and classify the
distances Euclidean distances between the points and
various parts of New York City. Since the industry is the corresponding centroid.
booming and expected to grow shortly. Effective taxi The important packages used in the project
dispatching will facilitate each driver and passenger to are pandas,numpy,seaborn,kmeans,yellowbrick and
reduce the wait time to seek out one another. The folium.
model is employed to predict the demand on points of
the city.
II. LITERATURE SURVEY
Keywords–Artificial Neural Network, Genetic
Algorithms, K-means Clustering, Recurrent Neural Past few years have seen tremendous
Network, Time delay Neural Network, Convolutional growth in uber related data analysis using machine
Neural Network.
learning. People are coming up with various methods
to analyze uber related data such as A state in which
I.INTRODUCTION the results, k-means clustering is used to estimate the
The Uber platform connects you with most likely collection points at a given time and to
drivers who can take you to your destination or predict the best hotspots of nightlife learning trends
location. This dataset includes primary data on Uber from previous Uber pickups. The center of the taxi
collections with details that include the date, time of service decides on the space of area to be targeted for
travel, as well as information on longitude and the pickup of passengers.
latitude in San Francisco and has operations in over
900 metropolitan areas worldwide. The prediction of This can be justified by explaining that
the frequency of trips of data is by implementing a machine learning is the core of Uber and how it has
part of k-means clustering algorithm impacted on tremendous growth
The standard algorithm describes the  Bridging the supply demand gap
maximum variance within the group as the number  Reduction in ETA
of square distances Euclidean distances between the  Route Optimization
points and the corresponding centroid.The use of the
digital computer has since moved to technology
where the program involves the use of neural
networks ,Examples of RNN (Recurrent Neural Poulsen, L.K In this document applied an
Network) and TDNN (Time delay Neural experiment of spatial analysis of Green cab and Uber
Network)for importing data from uber dataset which to hotspots of New York to determine the
takes the data for forecasting on a time horizon. competitive position of the NYCTLC. The resulted
research showed that as demand of green cabs on the complaints we demonstrate. The increased
hotspots grew,the demand of Uber taxis on the competition for these shared travel services has had
hotspots also growed. an intuitive impact on the behavior of taxi drivers[4].

This research recommends that NYCTLC Wallsten, S, stated that the results of New
creates a dashboard that analyzes and displays data York and Chicago are consistent with the possibility
in real time, as we believe this will increase its that taxis react to the new challenge by improving
competitiveness compared to Uber. Uber is a recent quality. In New York, the rise of Uber is linked to
taxi operator in New York and is constantly the reduction of objections to travel to the city. They
devouring the market share of the yellow and green discuss the competitive effect of sharing taxis in the
taxis of the New York Taxi and Limousine taxi industry using the complete data set of the New
Commission (NYCTLC).The NYCTLC is an agency York City Taxi and Limousine Commission for more
of the New York City Government which licenses than one billion taxi trips in complaints and details of
and regulates taxis and vehicle for hire industries and New York, New York and Chicago Google Trends
also app based companies. The commission was on the success of Uber's largest shared travel
founded on March 2 ,1971 and their headquarters are service.[5].
based in New York.[1].
Sotiropoulos, D.N, represented that this
Faghih, S.S recommends a recent modeling document addresses the problem of grouping, by
approach in Manhattan, New York City, to capture using a new approach to genetic algorithms that is
the demand for electronic mail services, particularly highly scalable in large volumes of textual details,
the Uber application. Uber collection data is added to developing a coding scheme based on centroids. We
the Manhattan TAD level and at 15-minute time apply k means clustering algorithm in this document.
intervals. This aggregation allows the Clustering is the unsupervised machine learning
implementation of a modern approach to spatio- algorithm used to solve grouping problems based on
temporal modeling to obtain a spatial and temporal similarities. This technique has aroused interest in a
understanding of the demand. During a typical day, wide range of scientific fields, which address
two spacetime models were developed using Uber clustering methods, to solve complex classification
collection data, the STAR and STAR and MSPE problems.[6].
turns determine the output of the models. The results
of the MSPE have shown that it is recommended to
Faghih, S.S said that the demand for
use the Lasso-Star system instead of the star design.
electronic mail services is growing rapidly,
A comparison between the demand for yellow and
particularly in large cities. Uber is the first and most
uber taxis in 2014 and 2015 in New York shows that
famous email company in the United States and New
the demand for uber has increased[2].
York City. A comparison between the demand for
yellow and Uber taxis in New York in 2014 and
Ghuhaexplained the grouping of the 2015 shows that the demand for Uber has increased.
sequences calculated and observed by using a small To study the forecast performance of the models,
amount of memory and time was necessary for you choose to choose data for a typical day. Our goal
applications that needed to develop a data flow in this document is to describe how these models can
model to involve large data sets and consider be used for forecasting Uber demand. The Uber data
categorizing the data in the form of clusters[3]. contains information about the position and time of
Ahmed, M., has shown that by using the pick-ups and returns of each trip during a day.
detailed data on taxis at the travel level and on the According to the available data, the Uber historical
rental vehicle and data on complaints about the level data of April 2014[7].Kumar, states that, k-means
of new complaints at the level of incidents, we study clustering is used to estimate the most likely
how Uber and Lyft enter damaged the quality of taxi collection points at a given time and to predict the
services in New York City. The overall effect of the best hotspots of nightlife learning trends from
organizations based on the scenario and in particular previous Uber pickups [8,9].
of the riding administrations was enormous and
widespread. One of these effects is the expansion of L.Liu, C.Andris, and C.Ratti planned for a
the rivalry between Uber and Lyft over the quality of strategy to disclose cabdrivers working patterns by
taxi administration. They use a new set of complaint inspecting their unbroken anatomy track[10].R-H
data to measure (the lack of) quality of service that Hwand focuses on GPS and the locality to pick up
we have never been analyzed before. Focus on the
quality dimensions generated by most of the
passengers , A venue to venue plot model referred to
as an OFF-ON model [11].

III. PROPOSED METHOD

Based on the problems of forecasting errors

and risk of overfitting due to large datasets. The data
analyzed and sent to the company is resulted as
inefficient and ineffective. Thus to overcome the
problem we are going to predict the pickup of cab
from a coordinated cluster of points predicted by
using applied k-means clustering algorithm.

The k-means clustering algorithm adopted

will effectively dispatch taxis to the cluster.This
facilitates each driver and passenger to attenuate the
wait-time to search out one another. Drivers don’t
have enough info concerning wherever passengers
and different taxis area unit and shall
move.Therefore, a cab center will organize the
taxicab fleet and with efficiency give out consistent
request to the whole town.

The system uses the latitude and longitude

of the cab scheduled and also the day of the travel
and the month.An unsupervised learning model is Fig 1. System Architecture
trained with this dataset and the model is employed
to predict the pickup of the cab on the cluster.The B. Raw Data(Dataset)
proposed method for the project is explained on 7
steps.
A. System Architecture The definition of raw data comes from the
concept of data not processed and is obtained from
B. Raw Data the dataset or is sometimes made by the end product
of data processing. The steps required in raw data are
C. Data Importing extraction, organization and sometimes analysis.
D. Data Visualization
E. Testing Data

F. Predicted Scheduling Of Cab Using

Algorithm
G. Algorithm

A. System Architecture
The system architecture for the given Fig 2. Raw dataset (csv file)
module is as follows:
C. Data Importing
A huge amount of trip data will be collected
fromUberfor training and testingdata. From the
collected dataset the latitude and latitude will be
clustered and classified based on the frequency of
trips travelled by the cab during the day. When these
criteria are considered, and data preprocess will be
done on these datasets.

Fig 5.Visualizing data on graph on frequency of trips

travelled during the day.

Fig 3. Processing the dataset

D. Data Visualization

Data visualization is defined as to evaluate

the performance of a model by using graphs and
metrics that calculate performance.Data visualization
can be mainly used to categorize the data into new
levels such that the algorithm used can be
generalized to an observation of each output variable
derived by an observed input variable.
Fig 6 Data Visualization of Heatmap of frequency of cabs
travelled during the hour.

E. Testing data

The main step after visualizing data in an

algorithm is to test the data,the test set can be
defined as a set of observations which is used to
evaluate the performance of a model by using
performance metrics.The program that uses the test
set must be able to generalize and effectively
perform with the dataset to yield the predicted data
accurately such that the program is effective in
nature. Moreover when the program memorizes the
dataset it is termed overfitting hence to balance
Fig 4. Data Visualization of the data based on graph overfitting we use regularization which is applied to
where graph has data of total trips travelled during the month .
the model to reduce it.

F. Predicted Scheduling of Cab using Algorithm

The scheduling of the cab can be predicted

on the basis of the location given by the user and the
proposed method finds the nearest hotspot which is
defined as a cluster of points analyzed by k-means
clustering and gives info to the cab on the hotspot
nearest to the location of the user and is booked to
pickup the user.
 grouping of k-means is to identify the clusters in
such a way to reduce total variation within the
cluster. The standard algorithm describes the
maximum variance within the group as the number
of square distances Euclidean distances between the
points and the corresponding centroid. The grouping
can be classified into two groups.

Hard grouping and soft grouping. Through

a hard grouping, each object or data point belongs to
a cluster. For example, all locations clustered in the
dataset belong to a district.In the soft grouping, a
data point can belong to more than one group with a
certain probability or probability value. In
connectivity-based clustering, the main idea behind
this cluster is that the data points closest to the data
space are more related than those of the data point
further back. Groups are created by linking data
points based on their length. Grouping based on
Fig.7.Visualization of total trips travelled by cabs on a day on the centroids, in this type of grouping, the groups are
basis of clusters. represented by a central vector or a centroid. This
G. Algorithm centroid may not necessarily be a member of the data
set. This is an iterative grouping algorithm in which
the notion of similarity derives from the proximity of
The algorithm used is involved on the the data point to the center of the cluster.
concept of k-means clustering algorithm. which
belongs to unsupervised learning. There is no labeled
data for this clustering, unlike in supervised learning.
K-Means performs division of objects into clusters
that share similarities and are dissimilar to the
objects belonging to another cluster.

Fig.8. K means clustering Fig.9 Importing k-means module and calculating the
centroids of the dataset

The clustering algorithm is categorized into

three steps:
 Take mean value
 Find nearest number of mean and
put in cluster
 Repeat the mean value and number
of means in the cluster till we get
same mean
Clustering is the method of grouping
objects into groups based on similarities. This
algorithm is used to divide a given data set into k
groupsHere, k represents the number of groups and
must be provided by the user. The idea behind the
 Fig 10. Plotting the centroids on x-y graph of latitude
and longitude.

IV. RESULTS AND DISCUSSION

The program predicts the pickup location of

the cab based on the centroids plotted using applied
by k-means clustering for appropriate cab scheduled
for pickup.The results discussed are based on the
following figures below

Fig.13. Labelling the dataset on the basis of clusters

Fig.11 Plotting the collection of points through which

cab has travelled during the course of the month.

Fig. 14. Predicting the pickup of the cab from the particular
cluster and showing the cluster coordinates.

V. CONCLUSION AND FUTURE WORK

The conclusion of the project is to project a

basic outline of trips travelled with respect to latitude
and longitude of locations and pinpoint the locations
travelled with respect to the frequency of trips
travelled by a uber cab during the day and also based
on the cross analyzing of the dataset based on the
latitude and longitude of the point travelled by the
cab which is then analyzed by deploying k-means
clustering which classifies the locations on the basis
of centroids and then orders the frequency of trips
Fig. 12. Plotting the centroids calculated by k-means
on the map of New York City imported by Folium.
based on labels or clusters. By the location given by
the user,the algorithm predicts the cluster nearest to
the location so that cab can be assigned to the user
for pickup.

The merits of the project is that it explains

the functioning of how cabs are assigned to
passengers based on an unsupervised algorithm and
also explains the key concepts of machine
learning.The limitations of the project are that the
algorithm deployed may be inefficient for huge data
for over 10 years.
 [9] Kumar, A., Surana, J., Kapoor, M. and
The future work suggests that the system Nahar, P.A., CSE 255 Assignment II
will provide the location to the user. The algorithm PerfectingPassenger Pickups: An Uber Case Study.
then records the time,latitude ,longitude of the trip [10] L.Liu, C.Andris, and C.Ratti , “Uncovering
and assigns it to a cluster nearest to the passenger cabdrivers behaviour patterns from their digital
location where a cab is scheduled for pickup. We can traces”,Compu.
also predict the passenger count on each district to Environ.UrbanSyst.,vol.34,no.6,pp.541-548,2010
deploy more cabs to the clustered coordinates using [11] R.H.Hwang,Y.L.Hsueh , and Y.T.Chen,”An
convolutional neural networks (CNN) effective taxi recommender system model on a
spatio-temporal factor analysis
model,”Inf.Sci.,vol.314,pp.28-40,2015.
REFERENCES [12] Vigneshwari, S., and M. Aramudhan. "Web
information extraction on multiple ontologies based
on concept relationships upon training the user
[1] Poulsen, L.K., Dekkers, D., Wagenaar, N., profiles." In Artificial Intelligence and Evolutionary
Snijders, W., Lewinsky, B., Mukkamala, Algorithms in Engineering Systems, pp. 1-8.
R.R.andVatrapu, R., 2016, June. Green Cabs vs. Springer, New Delhi, 2015.
Uber in New York City. In 2016 IEEEInternational
[13] L. Rayle, D. Dai, N. Chan, R. Cervero, and
Congress on Big Data (BigData Congress) (pp. 222-
S. Shaheen, “Just a better taxi? a survey-based
229). IEEE.
comparison of taxis, transit, and ridesourcing
[2] Faghih, S.S., Safikhani, A., Moghimi, B. services in san francisco,” Transport Policy, vol. 45,
and Kamga, C., 2017. Predicting Short-Term 01 2016.
UberDemand Using Spatio-Temporal Modeling: A
[14] O. Flores and L. Rayle, “How cities use
New York City Case Study. arXiv
regulation for innovation: the case of uber, lyft and
preprintarXiv:1712.02001.
sidecar in san francisco,” Transportation research
[3] Guha, S. and Mishra, N., 2016. Clustering procedia, vol. 25, pp. 3756–3768, 2017.
data streams. In Data stream management (pp.169-
[15] H. A. Chaudhari, J. W. Byers, and E. Terzi,
187). Springer, Berlin, Heidelberg.
“Putting data in the driver’s seat: Optimizing
[4] Ahmed, M., Johnson, E.B. and Kim, B.C., earnings for on-demand ride-hailing,” in Proceedings
2018. The Impact of Uber and Lyft on TaxiService of the Eleventh ACM International Conference on
Quality Evidence from New York City. Available at Web Search and Data Mining. ACM, 2018, pp. 90–
SSRN 3267082. 98.
[5] Wallsten, S., 2015. The competitive effects
of the sharing economy: how is Uber changingtaxis.
Technology Policy Institute, 22, pp.1-21.
[6] Sotiropoulos, D.N., Pournarakis, D.E. and
Giaglis, G.M., 2016, July. A genetic
algorithmapproach for topic clustering: A centroid-
based encoding scheme. In 2016 7th
InternationalConference on Information,
Intelligence, Systems & Applications (IISA) (pp. 1-
8). IEEE
[7] Faghih, S.S., Safikhani, A., Moghimi, B.
and Kamga, C., 2019. Predicting Short-TermUber
Demand in New York City Using Spatiotemporal
Modeling. Journal of Computing inCivil
Engineering, 33(3), p.05019002.
[8] Shah, D., Kumaran, A., Sen, R. and
Kumaraguru, P., 2019, May. Travel Time
EstimationAccuracy in Developing Regions: An
Empirical Case Study with Uber Data in Delhi-
NCR✱.In Companion Proceedings of The 2019
World Wide Web Conference (pp. 130-136). ACM.
C) SOURCE CODE

%pylab inline

import pandas

import seaborn

data=pandas.read_csv('Desktop/uber.csv')

data

data.tail()

data['Date/Time'] = data['Date/Time'].map(pandas.to_datetime)

def get_dom(dt):

return dt.day

data['dom'] = data['Date/Time'].map(get_dom)

data.tail()

def get_weekday(dt):

return dt.weekday()

data['weekday'] = data['Date/Time'].map(get_weekday)

def get_hour(dt):

return dt.hour

data['hour'] = data['Date/Time'].map(get_hour)

data.tail()

hist(data.dom, bins=30, rwidth=.8, range=(0.5, 30.5))

xlabel('date of the month')

ylabel('frequency')

title('Frequency by DoM - uber - Apr 2014')

def count_rows(rows):

return len(rows)

by_date = data.groupby('dom').apply(count_rows)

by_date

bar(range(1, 31), by_date)

by_date_sorted = by_date.sort_values()
by_date_sorted
bar(range(1, 31), by_date_sorted)
xticks(range(1,31), by_date_sorted.index)
xlabel('date of the month')
ylabel('frequency')
title('Frequency by DoM - uber - Apr 2014');
hist(data.hour, bins=24, range=(.5, 24))
hist(data.weekday, bins=7, range =(-.5,6.5), rwidth=.8, color='#AA6666', alpha=.4)
xticks(range(7), 'Mon Tue Wed Thu Fri Sat Sun'.split())
by_cross = data.groupby('weekday hour'.split()).apply(count_rows).unstack()
seaborn.heatmap (by_cross)
hist(data['Lat'], bins=100, range = (40.5, 41));
hist(data['Lon'], bins=100, range = (-74.1, -73.9));
hist(data['Lon'], bins=100, range = (-74.1, -73.9), color='g', alpha=.5, label = 'longitude')
grid()
legend(loc='upper left')
twiny()
hist(data['Lat'], bins=100, range = (40.5, 41), color='r', alpha=.5, label = 'latitude')
legend(loc='best');
figure(figsize=(20, 20))
plot(data['Lon'], data['Lat'], '.', ms=1, alpha=.5)
xlim(-74.2, -73.7) ylim(40.7, 41)
clus = data[['Lat','Lon']]
clus.dtypes
pip install sklearn
pip install yellowbrick
import matplotlib.pyplot as plt
from sklearn.cluster import KMeans
from yellowbrick.cluster import KElbowVisualizer
kmeans = KMeans(n_clusters = 5, random_state = 0)
kmeans.fit(clus)
centroids = kmeans.cluster_centers_
centroids
clocation = pandas.DataFrame(centroids, columns = ['Latitude', 'Longitude'])
clocation.head()
plt.scatter(clocation['Latitude'], clocation['Longitude'], marker = "x", color = 'R', s = 200)
pip install folium
import folium
centroid = clocation.values.tolist()
map = folium.Map(location = [40.71600413400166, -73.98971408426613], zoom_start = 10)
for point in range(0, len(centroid)):
folium.Marker(centroid[point], popup = centroid[point]).add_to(map)
map
label = kmeans.labels_
label
data_new = data
data_new['Clusters'] = label
data_new
seaborn.factorplot(data=data_new,x="Clusters",kind="count",size=7,aspect=2)
count_3 = 0
count_0 = 0
for value in data_new['Clusters']:
if value == 3:
count_3 += 1
if value == 0:
count_0 += 1
print(count_0, count_3)
new_location = [(40.86, -75.56)]
kmeans.predict(new_location)
clocation.head()