optimization problem - by navaneesh

25
Route Optimization Guide: Bhamati INSOFE – CPEE – Batch 14 G R Navaneesh Kumar

Upload: navaneesh-gangala

Post on 21-Jan-2018

244 views

Category:

Data & Analytics


1 download

TRANSCRIPT

Page 1: Optimization Problem - by Navaneesh

Route Optimization

Guide: Bhamati

INSOFE – CPEE – Batch 14

G R Navaneesh Kumar

Page 2: Optimization Problem - by Navaneesh

Problem Statement

Spring-Cleaning

Residential and Commercial Cleaning

Time – Resource – Money

Page 3: Optimization Problem - by Navaneesh

What I have

135 Min

45 Min

35 Min

25 Min35 Min

32 Min

55 Min

135 Min

Page 4: Optimization Problem - by Navaneesh

How I will tackle the problem

Optimize the Time

Travel Time

Execution Time

Save Time - Save Resource - Gain Profit

Page 5: Optimization Problem - by Navaneesh

Data – Available

JobExecutionTime.csv

Stop ID

Execution Time (Minutes)

TravelTime.csv

Stop ID 2

Stop ID 1

Travel Time (Minutes)

Page 6: Optimization Problem - by Navaneesh

Observations

1. 68 Locations Excluding Depot

2. Max execution Time :420Min

3. 75% Execution time <108 Min

Data Analysis-I

Page 7: Optimization Problem - by Navaneesh

Observations

1. 62 x 61 Travel Time – 3782 obs

2. Max Travel Time :135 Min

3. 75% Execution time <40 Min

4. Mostly Single City Works

Data Analysis-II

Page 8: Optimization Problem - by Navaneesh

Observations

1. 69 x 68 Travel Time – 4692 obs

2. Max Travel Time :135 Min

3. 50% Travel time <50 Min

4. Mostly Single City Works

Permutation of Data

Data Analysis-III

Page 9: Optimization Problem - by Navaneesh

Critical Business Conditions

Max Route Time < 720 min

Route Time = Travel Time + Execution Time

Max places vehicle can travel is 7

Excluding Depot it starts and stops

Locations it visits should be unique

No Location repeats in the travel

Page 10: Optimization Problem - by Navaneesh

D:X1:D - 68

D:X1:X2:D - 4556

D:X1:X2:X3:D - 3,00,696

D:X1:X2:X3:X4:D - 1,95,45,240

.

.

.

.

D:X1:X2…….X67:X68:D - 2.480035542 E+96

248003554243683059960099041

8569171581047399201355367672

3717107380182214457121832960

00000000000000

Challenge :

Looks Linear

Complex Solution Space

No Use of Linear Programming

Space and Time complexity issues

Business should not

stop for our model

Page 11: Optimization Problem - by Navaneesh

Genetic Algorithm:

Heuristic Method

Proved in Business, Science and Engineering

Best for searching for new solutions

Large population

Selection should be done

Survival of the fittest

Evolutionary Biology

Page 12: Optimization Problem - by Navaneesh

Basics of GA

Encoding principles (Gene, Chromosome)

Gene - Stop ID – Ex: Depot, 12350484, 12350757

Chromosome - Route - Ex: Depot : 12350484:Depot

Initialization procedure (Creation)

Creation - Initial Population

Ex: Depot : 12350484:Depot

Depot : 12350484: 12350757 :Depot

Depot : 12350484: 12350757 : 12350740 :Depot

Page 13: Optimization Problem - by Navaneesh

Selection of parents (Reproduction)

Here we use Business Constraints :

Ex: Stop ID (Gene) should not exceed 7

Each Stop ID should be visited only once

Total Time (Chromosome) =

Execution Time + Travel Time < 720 Min

Page 14: Optimization Problem - by Navaneesh

Genetic operators (Mutation, Crossover)

Crossover - Consider Two Chromosomes (Same Length)

Route 1: Depot : 12350444 :12350222 : 12350555 : 12350666 :Depot

Route 2: Depot : 12350888: 12350333 : 12350999 : 12350777 :Depot

12350444 :12350222 12350555 : 12350666

12350888: 12350333 12350999 : 12350777

Route 1: Depot : 12350999 :12350777 : 12350555 : 12350666 :Depot

Route 2: Depot : 12350888: 12350333 : 12350444 :12350222 :Depot

Page 15: Optimization Problem - by Navaneesh

Mutation

Route 1: Depot : 12350666 :12350222 : 12350555 : 12350444 :Depot

Depot 12350444 12350222 12350555 12350666 Depot

Route 1: Depot : 12350444 :12350222 : 12350555 : 12350666 :Depot

Combine

Page 16: Optimization Problem - by Navaneesh

Evaluation function (Fitness)

Termination condition

Below 75% of Data after Fitness Evaluation will be removed

Fitness = Fitness + Total Time

Total Time =Travel Time + Execution Time

Page 17: Optimization Problem - by Navaneesh

End

Start

Fitness assessment

BestChromosome

s to be selected

Selection

Crossover

Initial Population

Creation

New Population

Creation

Terminate

Condition

Yes

No

Page 18: Optimization Problem - by Navaneesh

ADVANTAGES OF GENETIC ALGORITHMS

A fastest search technique

GAs will produce "close" to optimal results in a

"reasonable" amount of time

Suitable for parallel processing

Fairly simple to develop

Makes no assumptions about the problem space

Page 19: Optimization Problem - by Navaneesh

Initial Population

Total Time ,Unique Ids

Page 20: Optimization Problem - by Navaneesh

First 1000 Iterations

Page 21: Optimization Problem - by Navaneesh

Next 10000 Iterations

Page 22: Optimization Problem - by Navaneesh

What I Achieved

Initial Time : 9113

Current Time :8048

Saved Time :1065

Nearly :17Hrs 45Min

A Nearly Two Working Days

Page 23: Optimization Problem - by Navaneesh

Future Scope

• Courier Services

• Food and Groceries Delivery App

• Container Shipping Industry

• And More

Page 24: Optimization Problem - by Navaneesh

Q and A

Page 25: Optimization Problem - by Navaneesh

Hasta la vista

-- Navaneesh Gangala