Webinar - Supply Chain Optimization: A Robust Supply That Minimizes Costs

Summary

本次演讲主要围绕供应链优化展开，介绍了一种旨在最小化成本并适应需求不确定性的鲁棒供应链模型。演讲者Fernando Piqueras，作为City of RAI的商业研究开发者，与FICO合作开发了一个决策支持系统，该系统使用FICO Express Insight平台进行库存管理。通过蒙特卡洛模拟和线性规划（MILP）技术，系统能够预测客户需求，优化库存水平，并在不同情况下做出最佳的订购和运输决策。此外，演讲还展示了如何通过用户友好的界面进行数据管理、模拟执行和结果分析，以及如何通过场景管理器快速比较不同策略的影响。整个演示强调了在不确定性环境下，如何利用先进的分析工具来提高供应链的灵活性和效率。

Takeaways

• 📈 **供应链优化目标**：创建一个能够应对需求不确定性的稳健供应链，以最小化成本并最大化需求满足率。
• 🤖 **决策支持系统**：开发了一个基于FICO Express Insight平台的决策支持系统，用于库存管理。
• 👨‍💼 **演讲者背景**：Fernando Piqueras作为业务研究开发人员，拥有在零售、运输和物流行业的丰富经验。
• 📚 **合作与技术**：与FICO合作，利用其领先的分析解决方案，通过操作研究技术提供定制解决方案。
• 🔄 **灵活性与敏捷性**：平台能够快速适应变化，为业务提供最佳策略，并加速实现业务目标。
• 📊 **多场景模拟**：通过蒙特卡洛模拟，能够可视化多种可能的业务结果，帮助分析师和业务用户进行决策。
• 📦 **库存管理决策**：需要决定向供应商订购的产品数量以及从哪个仓库向客户交付产品。
• 📈 **成本效益分析**：考虑了最小化库存成本和分销成本，包括订单处理和产品存储成本。
• 🗺️ **地理分布考虑**：模型考虑了地理位置，以确定最短距离和最低成本的配送路线。
• 📊 **关键绩效指标(KPIs)**：通过定义KPIs来衡量解决方案的表现，包括成本和服务水平协议(SLA)。
• 🔧 **实时模拟与分析**：应用允许用户进行实时模拟，观察不同决策对供应链和成本的影响。

Q & A

什么是供应链优化的主要目标？

供应链优化的主要目标是创建一个能够应对需求不确定性的稳健供应链，以最大化需求满足率，同时最小化库存成本和分销成本。

Fernando Piqueras 在哪个领域工作，并且他的主要工作内容是什么？

Fernando Piqueras 是一位商业研究开发者，他在 City of RAI 工作，专注于使用运筹学技术为客户创建定制解决方案，涉及零售、运输和物流等多个项目。

FICO 在供应链优化中扮演了什么角色？

FICO 是世界领先的分析解决方案和服务提供商，它通过其平台帮助组织更好地理解客户，专注于优化业务流程。在供应链优化中，FICO 提供了技术支持，使得项目能够快速扩展并迅速应用于业务。

在供应链优化中，为什么需要一个决策支持系统？

决策支持系统对于库存管理至关重要，因为它可以帮助应对需求的不确定性，提供基于数据的决策支持，从而提高供应链的灵活性和响应能力。

在供应链优化问题中，有哪些关键的决策变量？

关键的决策变量包括向供应商下达的订单量、仓库的存储量、从哪个仓库向哪个客户配送产品，以及如何处理产品从供应商到仓库、从仓库到客户的运输。

如何使用 FICO 平台来开发供应链优化的解决方案？

使用 FICO 平台，可以在没有先验知识的情况下从零开始开发解决方案。通过 FICO Express Optimization 软件，可以解决和优化混合整数线性规划（MILP）模型，同时 FICO Express Insight 提供了用户界面，允许用户管理数据并可视化可能的业务场景。

在供应链优化中，如何处理地理分布对成本和配送时间的影响？

在供应链优化模型中，需要考虑供应商、仓库和客户之间的地理分布，以确定最短的运输距离，从而最小化运输成本和配送时间。

在供应链优化的第一阶段和第二阶段中，分别需要做出哪些决策？

第一阶段主要是根据需求预测向供应商下达订单，决定每个仓库在特定时间内对特定产品的订单量。第二阶段是在已知实际需求后，决定从哪个仓库向哪个客户配送多少产品。

如何通过 FICO Express Insight 平台进行实时供应链优化演示？

通过 FICO Express Insight 平台，用户可以登录到服务器，通过不同的视图（如输入数据视图、输出结果视图、分析视图）来管理数据、执行模型、查看结果，并进行实时演示。平台提供了交互式的用户界面，允许用户加载数据、运行模拟、查看订单和配送情况，并对不同场景进行分析。

在供应链优化的实时演示中，如何评估解决方案的效果？

通过关键绩效指标（KPIs）来评估解决方案的效果。这些 KPIs 包括总成本（如订单成本、配送成本、库存成本）、服务水平协议（SLA，即满足需求的比例）等。用户可以在 FICO Express Insight 平台上查看这些 KPIs，并根据需要自定义它们。

在供应链优化中，如何考虑不同的业务场景和假设情况？

通过创建不同的场景，用户可以模拟不同的业务场景和假设情况，如不同的需求预测、仓库配置变化或地理分布调整。FICO 平台允许用户快速修改参数并重新运行模型，以评估不同决策对供应链优化的影响。

Outlines

00:00

😀 欢迎与介绍

演讲者Fernando Piqueras介绍了自己作为商业研究开发者的身份，并简述了其在City AI的工作内容，包括与FICO合作开发基于运营研究技术的定制解决方案。强调了FICO在分析解决方案和服务方面的领导地位，以及如何通过优化和业务流程改进帮助组织更好地理解客户。

05:03

📈 供应链优化问题定义

讨论了供应链优化的目标，包括最大化需求满足率、最小化库存成本和分销成本。介绍了决策变量，如向供应商下单和从哪个仓库向客户配送产品。同时，考虑了关键因素，如产品地理分布、最短距离和成本。

10:05

🛠️ 解决方案构建与模型介绍

详细说明了如何构建解决方案，包括使用FICO平台开发应用程序的过程。介绍了模型的两个阶段：第一阶段是提前下单，第二阶段是使用这些订单满足客户需求。还介绍了如何使用FICO Express软件来解决混合整数线性规划（MILP）问题。

15:05

📊 数据输入与模拟

展示了如何加载和修改数据，包括需求预测和蒙特卡洛模拟。讨论了如何通过用户界面与应用程序交互，并展示了不同视图，如输入、输出和分析视图。

20:07

🏭 仓库配置与地理分布

介绍了如何配置仓库、供应商和客户数据，以及如何在地图视图中查看这些数据的地理分布。讨论了如何通过模拟来预测和规划仓库的订单和客户的配送。

25:10

📊 数据输出与执行

展示了如何查看模型的输出，包括订单和配送数据。讨论了如何通过地图视图和表格来理解这些数据，并如何通过执行不同的模拟来评估解决方案。

30:10

📊 库存水平与KPI分析

介绍了如何查看每个仓库的库存水平，并使用关键绩效指标（KPI）来分析解决方案的效果。讨论了如何通过图表和表格来展示总成本、服务水平协议（SLA）和其他相关数据。

35:12

🤔 情景分析与决策

讨论了如何通过创建和比较不同的假设情景来评估不同决策的影响。展示了如何使用情景管理器来添加和修改情景，并如何通过分析不同情景的KPI来做出决策。

40:13

📝 总结与联系信息

总结了演讲的主要内容，包括供应链优化的必要性、提出的解决方案、用户友好的界面以及如何通过KPI进行问题和解决方案的深入分析。最后，提供了联系信息以便进一步的交流和问题解答。

Keywords

💡供应链优化

供应链优化是指通过改进供应链流程和物流，以减少成本、提高效率和响应市场变化的能力。在视频中，供应链优化是核心主题，涉及到通过一个决策支持系统来管理库存，以应对需求的不确定性。

💡成本最小化

成本最小化是企业在追求经济效益时的一个关键目标，它涉及到减少生产、运营等过程中的各种成本。视频中提到，通过建立一个健壮的供应链，旨在实现成本最小化，同时满足需求不确定性。

💡需求不确定性

需求不确定性是指产品或服务的市场需求在未来的变化是不可预测的。视频中强调了需求不确定性对供应链的影响，以及如何通过优化策略来应对这种不确定性。

💡决策支持系统

决策支持系统是一种帮助管理者通过数据和模型来做出更明智决策的工具。在视频中，开发了一个决策支持系统，用于库存管理，以便更好地应对需求波动和优化供应链。

💡FICO

FICO是一家提供分析解决方案和服务的领先供应商，帮助组织更好地理解他们的客户并优化业务流程。视频中提到与FICO的合作，利用其领先的分析技术来支持供应链优化项目。

💡蒙特卡洛模拟

蒙特卡洛模拟是一种数学技术，用于通过模拟随机变量的重复抽样来估计可能的结果分布。视频中使用蒙特卡洛模拟来预测客户需求的可能变化，这是优化库存和供应链管理的关键步骤。

💡关键绩效指标(KPI)

关键绩效指标(KPI)是衡量组织、团队或个人在实现目标方面表现的量化指标。视频中提到使用KPI来评估供应链优化解决方案的效果，如成本和服务水平协议的达成情况。

💡地理分布

地理分布涉及到地理位置的布局，对于供应链管理来说，考虑产品的地理位置对于优化运输路线和降低成本至关重要。视频中讨论了如何通过考虑地理分布来优化供应链。

💡库存成本

库存成本包括持有库存商品所涉及的所有费用，如存储费用、保险费用、折旧等。视频中提到了最小化库存成本作为供应链优化的一个目标，以减少整体的运营成本。

💡服务水平协议(SLA)

服务水平协议(SLA)是服务提供者和客户之间关于服务质量的正式文档。在视频中，SLA用于衡量供应链解决方案满足客户需求的能力，是评估供应链性能的一个重要指标。

💡模拟

模拟是指创建一个模型或环境来模仿真实世界的情况，以便进行测试和分析。视频中提到了使用模拟来预测和准备不同的供应链场景，如仓库不可用时的应对策略。

Highlights

演讲主题为供应链优化，旨在创建一个能够应对需求不确定性的鲁棒型供应链，以降低成本并提高效率。

开发了一个决策支持系统，使用FICO Express Insight平台进行库存管理。

演讲者Fernando Piqueras是City of RAI的商业研究开发者，专注于基于运筹学技术的客户定制解决方案。

与FICO合作，FICO是分析解决方案和服务的世界领先提供商，帮助组织优化业务流程。

提出的解决方案能够更敏捷地应对变化，通过演示展示如何自信地部署最佳策略。

平台能够加速实现业务目标，通过物化业务概念来加速达成可扩展性和采用性。

应用从零开始开发，仅用三周时间，展示了FICO平台的高效性。

应用允许用户可视化多种业务情景，为分析师和业务用户提供合作空间。

供应链模型考虑了客户、产品、供应商、仓库和地理分布等多个因素。

通过蒙特卡洛模拟，基于需求预测创建可能的客户需求场景。

解决方案包括两个阶段：第一阶段是确定订单，第二阶段是基于实际需求确定发货。

使用FICO Express Optimization软件来解决混合整数线性规划（MILP）模型。

应用程序界面允许用户管理数据并执行模型，同时数据托管在服务器上。

演示展示了如何使用应用程序进行日常操作，包括需求预测、模拟、配置和执行模型。

通过地图视图，用户可以直观地看到供应链中不同组件的地理分布。

输出视图提供了订单和交付的详细信息，帮助用户了解库存水平和供应链状态。

KPI分析视图提供了关键性能指标，帮助评估解决方案的效率和成本。

通过场景管理器，用户可以创建和比较不同的供应链情景，以适应不同的业务需求。

演讲总结了供应链优化应用的关键特点，包括灵活性、库存管理、用户友好的界面和情景分析。

Transcripts

hello everyone welcome good morning

uh we're gonna be doing a presentation

on supply chain optimization

a robust supply that minimizes cost uh

the idea

here is to make a supply that is able to

cover

uncertainty on our demand that's why it

needs to be robust

um what we just developed is a decision

support system

uh for stock management that we've used

that using the flag express insight

platform we're gonna see more into

detail on

everything here in the screen in a

little bit first off

uh my name is fernando piqueras i'm a

business research developer

at the city of rai uh on the screen um

we have some contact uh information you

have a webpage as well as our emails

uh you can shoot any questions or doubts

uh we'll

gladly address those uh as for us well

uh i've been working at the city for

like two years

uh here at the city for ai we create

taylor solutions

based on operations research techniques

of just

to cover our customer needs i've been

involved

in many multiple projects mostly in

retail

transport and logistics sectors

so we've worked in collaboration with

fico fico is the world leading

provider of analytics so world solutions

and services

it just transforms the way organizations

understand

their customers and be focused on

optimization and business processes

around them so some key fico takeaways

that we're going to see today

uh both uh bio to business and then how

it

scales rapidly and the quicker option

into the business the projects

uh so first off uh the value is going to

see that uh we're going to be able to

react to change with more agility i hope

that the

demo that we're gonna do at the end of

the presentation and cover this uh

the idea is to deploy the best champion

strategy

with confidence uh another key takeaway

from fico is

how the platform is going to accelerate

the realization of business objectives

we're going to objectivize

our these business concepts and then

we're going to

accelerate towards like meeting those in

terms of

scalability and adoption uh well first

off to start the application you're

gonna see today

has been developed from scratch without

any knowledge prior knowledge

uh using the fico platform we've

developed that within three weeks

so i think it's pretty telling we have

these key takeaways uh we're going to be

able to visualize possible multi

multiple scenarios for our business

outcomes

i think this is a very key important

part we're going to see at the end uh

having multiple scenarios uh also

enables collaborations between

analysts and users in business terms

the application is made for both uh

users and analysts

and we're gonna see less dependence on

on other it

this is the outline of today's

presentation uh first off we're gonna

start with some context

an introduction and definition problem

exactly what our plumbing needs

or what is exactly what we're addressing

today

uh afterwards uh we're gonna go more

into detail about the problem statement

the considerations we've taken on our

project

uh and finally um our milk solution

uh how we've formulated our milk problem

um once we have all the all the

ideas clear about what the problem is

about

uh this is stock optimization problem uh

we're gonna go into

the methods and technology used on our

application we're gonna go into

the architecture of ficus process

optimization that we've used on our

application

and finally we're going to do a live

demo of well

the application that we've developed

so to start with the context the problem

is going to be set on

a supply chain right let us assume a

supply chain uh composed by first

a set of customers uh each period

uh these customers are going to demand

an uncertain quantity of each product

right the key concepts that is that

quantity is uncertain

at the beginning um in order to meet the

customer demand we need

warehouses that are going to be able to

store the products

and of that there's another layer of set

of suppliers

that will supply these products to

warehouses this would be a normal

uh supply chain

so our problem definition um

well the first is that you have a few

objectives the most important objective

to me

is that we're gonna be able to maximize

this demand fulfillment we want to be

able to meet

uh as many demands as we can this is our

key product um

key objective the most important one but

as well we have other objectives

to meet along the way uh such as uh we

want to minimize

the stocking costs that warehouses uh in

order to supply this uncertainty in

demand we're gonna be able to

we're gonna have to stop products in our

warehouses we're going to minimize the

product

the cost of those as well as we want to

minimize

the distribution cost both for the

orders to suppliers

and as well to the deliveries to the

customers

the decisions that we're gonna make is

well i've already introduced them

we're gonna have to uh make a decision

on the orders to suppliers and also

which from our warehouses from what from

which of our warehouses we're gonna

deliver to which customers

um the key components that we have to

consider on these decisions is that well

you have to address uh where this is

where these decisions can be made

um how much of each product uh what what

kind of product

and as well i have to um deal

with the from where to where uh

schematics we're gonna

model a project on a with considering

geographical distribution we have to

consider

which ones are the shortest distances uh

to for all these deliverers

okay uh our problem statement these are

the considerations that we've taken

our model the project uh the problem

let's see these are the indexes that i

have right we have a set of periods

uh for instance weeks we've used a set

of products

uh suppliers warehouses and customers

these are the

the data we're gonna see on our model

so first off our considerations for

suppliers i've considered

that when making orders to suppliers

there's going to be a minimum

a maximum quantity that you can order at

once

as well as on our warehouses we're going

to have a maximum storage capacity

we cannot store more than that on our

warehouses

and then all the products that are being

stocked there's gonna be a unitary

storage cost for those products

uh considering our customers we're gonna

have uh

an expected demand uh obviously it's not

final but it's

a sort of like a forecaster more into

detail later

and then we have an undeliver cost a

penalty for not being able to

meet the demand

for as well as our considerations in the

actual orders to the suppliers

uh we have a fixed cost per order uh

which makes sense and then as well

uh the variable cost component the cost

per unit order

we also have to consider a lit time uh

the time that it takes um

for the product uh from for the order

from the supplier to arrive at the

warehouse

and all these considerations are

obviously dependent on the distance

between the supply and the warehouse

on the other side for our deliveries to

customers we have

pretty much a very similar set of

considerations

i have fixed cost per delivery i have a

unitary cost per delivery we have both

the fix and the variable costs of those

deliverers

as well as and again another delivery

time and all this is based also again

in between the distance uh from the

warehouse to the customer

these are all the considerations that

we're gonna see and we have to model

a lot of this into our problem so we've

decided we've proposed

a build model to solve the previous

explain problem

so we're going to start with our initial

inputs our data we're gonna have a

demand forecast

uh demand forecast is gonna tell us uh

warehouses how much of each product they

need

it's gonna abstract the concept of

customers

we're not going to be able to see the

demands at that point which is going to

have a forecast

and with the forecast we're going to do

monte carlo simulations

to simulate what would possible demands

for customers

look like so we're going to have to take

uh we're going to take this

forecast model as our input together

with the problem data this problem

variable

the geographical distribution or where

everything is situated

as well as this fixed and variable cost

that we've introduced

this will be our inputs for the model

are we going to feed this into the milk

um i'm going to explain in the next

slides how this milk is composed of

but just to introduce we're going to

divide this into two different stages

um and then what we're going to obtain

out of this milk or

this black box of a milk uh it's going

to be our movements in the warehouses as

we

explained before these decisions that

we've taken

uh so go more into detail on each part

first off uh for the first stage of the

milp

are we gonna be working with this

decision variable the order variable

this is gonna be the quantity order

by a specific warehouse of a product p

uh to a supplier in a period of time

so again the input data at this part is

this demand forecast that we talked

about

uh we're going to fit this into the milp

stage and we're going to obtain

this decision variable so the key

concept is that at the beginning on our

first stage

we do not know our exact demand so we

cannot plan around that

however we're going to use this forecast

to create our set of orders that we're

going to make

the idea is that we have to order in

advance

at this first stage and then in the

second stage it's when we're gonna

uh use these orders to supply our

customers so in the second stage of the

problem

uh we're gonna work with obtaining the

second decision variable the shipment

one

this time for a specific period we're

gonna this the variable is gonna say how

much quantity is even shipped

from a warehouse of a display product to

a customer

so again we have this output from phase

one that was the order decision variable

we can now have it as an input for the

second stage

and we're gonna put it together with the

actual real demand the expected demand

after we're made of orders in advance

now we actually do

know what our demands are gonna be like

and we have to

supply these demands right so we're

going to fit this into our middle stage

and we're going to obtain the actual

real shipments on our

on our model so you look at it together

but a holistic point of view you have

these two decision variables the order

and the shipments

you're going to fit as expected demand

together with the plan data

and we're going to obtain in general

these two decisions variables

so methods and technology we've

developed an application um

the applications using the fico express

sur

software so underlying all of the

application to solve this milp

you have an optimization software if you

use the fico express optimization

this is the one the the software that's

going to be solving and optimizing

the milk model that we just explained on

top of it

this application is going to allow the

users to manage the data

such as our suppliers our warehouses and

our customers

on top of that the user is going to be

able to inter uh

interact with application through a user

interface this is being done

through the fico express insight so far

the demo that we're going to see today

is going to be on this platform on the

final express incident platform

and then our data is going to be hosted

on the server data on the inside data

server

this is a look of how our application is

going to look like

now we're going to use a live demo we're

going to use the demonstration of house

our day-to-day operations

um with this application the solution

that we've proposed

so for our demo we're going to have

three different stages

uh from previous demonstration first of

all we're gonna look at the first stage

how we dealt with this robust stock

after we've done

this robust stock we're gonna work into

how the supply chain has been addressed

and finally we're gonna see some results

analysis so without any more

further ado let's jump into the

application

uh this we are already logged in to the

server

uh hosted on fico in server file express

insight

this is what you will see when you log

in at first i have we have the name of

the application at the top

and we already selected the app right

this right here this bar is a scenario

manager

more into that later but what we would

see at first

is these different tabs that they all

express

different views the views are obviously

what we see on screen

so if we click on the on on this

menu bar we see the view navigator that

contains all the views of our

application

we have a set of views as inputs uh with

different inputs that we consider for

our problem we got a set of views

for the outputs as well as our set of

views for the analysis

so we're now in the input section and

these are our four different

input tabs the one that we're seeing

right now on the screen

is the is the forecast and simulations

uh view

and here we're gonna see our demand

forecasting

and as well as the simulations the monte

carlo simulations

for the demand based on this forecast we

have another

ebu for the warehouse configuration and

here we have the rest of the input data

for our problem related to warehouses

and suppliers and we've been able to

configure

some parameter settings for the

execution of our model

we have a demand view that we're not

going to see it at this first stage

this one is for the second stage of the

problem this is where we would actually

introduce

the real demand the final demand that we

have to supply

and we also have a bump view we're going

to be able to see these inputs

uh in a map view so what we're seeing on

this screen

uh we're seeing actual data it's already

loaded

right so we've done so by using the

scenario manager at the top

if we click on the scenario manager

these are all the different scenarios we

can consider scenarios as data sets

so we have one already scenario loaded

in this column on the shelf this one is

ticked the shelf is

the scenarios that we already have here

added to them

so we want to add more scenarios to our

self then we just click on those

and we're going to press the add button

and we see that

they're added here at uh our cell to

ourself

so for today's demo we have our company

plan already made

uh we have different scenarios for the

data at every single stage

so we're gonna add all three of our

companies plan

to ourselves the day if we close the

scenario manager

the data that we're seeing uh on screen

on our views is the one with the

scenario that is farther on the left we

can move this around we're going to do

that later to change scenarios but the

one we're seeing the data

it's from this one the left outermost so

they actually have in this

view we have first our forecast table

uh this the forecast table again for

a warehouse it will tell us how much

quantity is x forecasted to be needed

for a specific product at that specific

period

this will be our starting base on we're

going to

work with so this is obviously pretty

extensive we have

just you can only see some of the period

one but you can obviously

move around the table with this arrows

uh you can see here some of the period

two as well

so if we take a look at uh the the

quantity uh

we're gonna see later how this data is

loaded but know that we can actually

modify this

uh easily if you're gonna change around

and play with our forecast

to see what like hypothetic scenarios

where i meet might have

more or less forecast uh

how would i deal with that can figure

out some few buttons

to modify the forecast on the table you

can add a value

here a percent to a value like 20 and we

can press the button for instance to

increase the forecast we're going to see

that when this action is being executed

by clicking it

our forecast has gone up by 20

the table is really good to have all the

information that we need

uh basically i import from a spreadsheet

right

but for us humans it's way easier to

see these values here on our forecast

graph

so in this forecast graph you're going

to have the same information a table

however we're going to be able to see it

uh again graphically so our graph has

the periods

on our x-axis we're gonna see all the

different periods

as well on the y-axis you can see the

quantity for each product

and the way we have each product here

addressed

we have your legend of the different

product units right we have sq

sq sku from one to six

and we can see the different products by

hovering over the data

if we have the mouse over one we can see

the different

uh values for each individual uh color

coded by

this legend right so for instance on

this

on this graph you're able to be telling

more detailed info

on our forecast situation uh we can

easily

see that our most busier periods are

gonna be period four

five and six being five our highest uh

demand

uh period based on this forecast

obviously whereas uh period one

and period eight are gonna be less

intensive

compared to the other ones this is

something that we can easily see

and understand based on looking at the

graph

then again from this forecast we've

commented

that we've made several scenarios okay

so we have here five different

simulations

okay and these simulations are just

monte carlo we're gonna be

create uh fake demands for our customers

based on this simulation so we click on

one

uh they're gonna see the demand

table populated you can see how our

simulations look like

for each of the simulations here on this

uh drop down

so on this demand table we have the data

my id the expected quantity

for a specific period for a customer

and a for a type of product obviously a

daily simulation and we also have as

well

the penalty the cost for not being able

to deliver

uh this demand that is currently set to

the same for all of them

but it would allow to give different

priorities to demands

again this information has like 26 pages

of data

however we can easily address this

the demand and we can see

and have a better look into it using

this

uh graph again we have all this

information on the table

but we can see uh the data here more

clear

you can understand that it has the same

tendency as before for the forecast

where period 5 was the busiest however

it's

not as it's obviously random so the data

is skewed

subjected to randomness obviously this

would be

how we would use some data uh

in the five expressions application

we've seen how

the the inputs for our forecast and

simulations in our application we're

gonna go now into the next view

for the warehouse configuration well

again in this step

the first buttons we're gonna see we're

gonna see the load button

obviously we're seeing all the data

populated

on our scenarios up here however this is

being done

uh by loading the data first off so how

this is being done

if you go to the top right here to the

gear for the settings for the actions if

you click on it

you're gonna see here app attachments if

you click on app attachments

you're going to see a set of csv files

that we've uploaded here

and these ones have the information that

we see on the table so like a user may

upload their own files here

and then when clicking the load button

all of this data will be loaded into the

tables

uh this bottom for run simulation is

going to execute the first stage of the

model

we're going to see that later when we uh

actually execute the pro

the model however looking into what we

have here

you have the rest of the tables for our

input data

we have a warehouse table we're going to

see all of our different warehouses we

currently have three warehouses

we see the stocking cost as well as the

maximum capacity

of storage for for units you can as well

set up a warehouse to be unavailable

for the execution um of the model

as in we're gonna see that later in

alternate scenario would happen if we

just

do not count with that warehouse the

data here

is able to be edited you can mark it or

mark it as

unavailable as well as we could edit

these values these numeric values we

could probably like increase the maximum

stock

i'm going to set it back to where it was

this is making reference to the

warehouse configuration

idea of this view next off we have the

suppliers table we have our set of

suppliers

again we have a fixed delivery cost for

making these orders as well as we could

also set the suppliers as being

unavailable

uh for the model execution as well as we

have the rest of the data

we have a different set of periods where

we have to consider these ones

uh for instance is weeks they're just

like example periods

then you have our customer table with

the latitude and longitude

of where these customers are located as

well as our different product table

um in order to be able to see

more into detail on how

my geographical distribution of my

customers my warehouses and suppliers

we're gonna go into the next view again

we're gonna jump over

this next view to the demand view we're

going to see that on the second face for

the real demand we're going to go into

the last view

of the inputs to the map view and when

we go into this map view

we're going to see in here um

we're going to see this like blue

brown and red this is going to be our

our different warehouses

okay uh whereas our

our green triangles are going to be our

suppliers if we can hover over this we

can

it's going to tell us information from

our suppliers

uh finally we have the different dots

these are our customers

our sort of customers so here in this

map you can have a better look

at what our dataset looks like

on the map view

so this has been

this has been all part of the inputs now

what we do is we could actually run the

simulations to run the first stage of

the model

we've already done so and we have the

data we run

uh on this other scenario on the second

stage

we've done so just priority to avoid uh

the waiting time so we're gonna

move it the scenario by drag it to the

first position

and when we do so it's gonna change the

data uh the scenario is mostly the same

except it's already been run

so if we click to the abuse and we go to

the outputs view

this output view we're gonna focus in on

these orders and delivers

on this view we're gonna see both the

orders and deliveries our

most important identifier to see at this

point

is that this table of orders has been

populated

these will be the orders that we're

gonna obtain

from this first stage we had this insert

in demand and we wanted to

take decisions on which which of our

orders are going to be

to populate our warehouses with stock

we obviously use the simulations uh to

be able to simulate later deliveries

to have a full product but here we can

see on the table

that we have all for each period we have

the different suppliers that are making

uh supplies of different products and we

have here the quantity column

and to which warehouses are being

delivered

here at the bottom we would have a

different drop down menu to check all

these different simulations

and then we have the custom deliveries

for the the demands on the simulations

uh this is only this is not a real

output

of the problem because we're going to

use the second stage to obtain the real

deliveries however it's

it's it's it's here to evaluate

uh how it might have reacted to

different simulations if you wish to

see that um so this has been the first

stage

of our live demo we've looked into the

demand forecast

uh how we handle these data inputs how

is the forecast being treated

as well as the simulations being done

from the forecast

we've looked also into the work

configurations view uh we were able to

mod a little bit how the problem data is

going to be as well as take a look at

the rest of the inputs

we've seen this problem data in the map

view uh we could

locate every single component

geographically

as well as after doing the first

execution we were able to see the orders

output this was the key takeaway from

the first stage

now we're going to go into the second

stage of the supply chain

so once uh we already have the

the orders right we're going to go back

into the input so i'm going to go to

this view that we skipped to the demand

view

we're going to upload again through the

load uh

a set of data with the final demand and

then this is the page we would work with

the real demand so first off we have two

different buttons

for execution modes we have a first

button for the complete run

if you wanted to just execute it faster

this bottom would execute both

the first and the second stage provided

we have all the data already

added to it both the forecast and the

real demand

and then as well we also have a button

to run the file demand this would run

the second

stage of the execution the one that we

currently add

so you have again a demand table this

demand table

has the demand for the finance

simulation the actual real data so these

are the real expected quantities

that we need and then again we can take

a look at this data

in a graph here to be able to see a

iso breakdown of the different products

the way we've obtained this

data uh we did five simulate monte carlo

simulations for us

uh to obtain our first five simulations

have done a sixth

one and this sixth one is not being used

for the first stage but it's now being

added

to the to the model to work with it now

so what we will do now

is you would run the final demand uh

button to execute the second stage of

the problem

we've already done so and we have the

data on this last

scenario on the company's plan complete

so we're going to drag this scenario to

the beginning

and then we're going to explore uh

different datas so if you go to orders

and deliveries

back at here we have now the final

simulation selected

and we're gonna be able to see all the

custom deliveries

for this um for this customer delivers

for this final simulation

uh again this this this is very

extensive to see

on the table there's a lot of data for

uh because of our demo data

however we can see this we're gonna skip

for a little while the stock view

we're gonna go into the map view and we

can see all these deliveries being done

over here so we had again our three

different

warehouses right if we click on those

and if we click on a warehouse you can

see all the different shipments that

we've made

um you can see this uh

through these lines here on the map uh

we've assumed

it you can see that from warehouse three

this one

we're supplying to a lot of customers as

well as also supplying a few others from

like warehouse two

um this is the uh an interactable view

where we can see

our data that we had on our table we can

see

our orders here

from the work from the from the

suppliers and we can see also

the deliveries from the warehouses

a quick easy view to understand all the

information

that we had on the table the output of

our solution

if you go back to to the stock view

this is the other second output that

we're going to have

in this view you're gonna be able to see

the stock levels for each warehouse

uh for different simulations so we're

gonna pick uh warehouse three here

i'm gonna select the final simulation

and we're gonna see if we have a table

uh on this table we have as rows the

different products right we have the sq

sq one two six as well as the final row

with the all this is the

sum of all the rest and you have per

columns you're gonna have the different

periods

the quantity of the the stock levels for

each different period

it's from period zero to period eight

and then well the the actual quantity

value in the table on that on the

on that matrix uh it's easier to digest

all this information

if you have a look at it on the graph

right this is the graph uh that we're

going to have the different periods

and the quantity for each product so

we're going to see the stock

period and product however we've added

this

sq all the cemetery of all the products

uh that is um altering our

different uh edges like we it's mostly

we can in the in this graph you can

barely see the other products because

the sum

is quite higher than the all the others

so we're gonna see we're gonna go back

to this legend view

i'm gonna click on the sq all so we can

disable it from being

unit on the graph and now we can

actually have a better view

of the individual products uh per period

so we see that we stuck

early on these early early periods

right uh having this like middle periods

specifically period four like our

busiest um

fullest and then for the last period we

were like released all the stock to

handle all this supply

uh this would be how we view uh the

different

stock levels for our warehouses on our

current solution

so now that we've seen all the outputs

for the data the most sensitive part is

to

uh analyze how this solution was so

we're gonna go into our views

we're gonna go into analysis we're going

to go to the kpis view

okay the scenarios will see it later but

for now we're going to work on the kpis

view so we already have

um selected a simulation in this view

you're going to see the kps for the

current selected simulation

we're going to use the real data the

final one and we're going to see in here

are the different measurements our key

performance indicators that we

have decided to show on this application

these are the data that we think

they think that it's going to be able to

tell how good the solution was

a comparison these are the value these

are the two kpis

that we've decided to show this

obviously would be

adapted to our customer needs uh

depending on

the actual context of the problem and

this can be easily customized to any

customers so we have two different ones

we have one for cost

and the other one for the service level

agreement so looking at the cost

we have first a table for the total

costs we have for the simulation

selected the order cost

uh this is for the orders from suppliers

to warehouses

the shipment cost this is the cost from

the deliveries from

warehouses to customers as well as the

stocking cost this is the cost for

having the units

uh being stocked in the warehouse and

you have a final row

with the total cost the sum of all the

costs

we can see these different stocking and

distribution cost on this graph

right here it's a stock of our stacked

bars graph

we can have a different color code for

all the different costs and we

hover over each of the periods that we

have as our x axis we can see

the different breakdown of this cost

where we mostly on our solution

uh we have a lot of different shipments

uh which are the

the busiest part or the ones with higher

cost

in general uh we also have this

uh next table we have a breakdown of the

cost per period

we have both uh for a selected

simulation and for

all different periods we have the

stocking cost right the

the custom warehouses as well as the

distribution cost this distribution cost

is the sum of both the orders and the

shipments

this would be our first kpi uh different

the the cost of the current solution

uh as our second kpi we have the service

level agreements

based on our demands how many are we

able to meet and how many are we not

able to meet

so we have an sla table right for our

current simulation and for different

periods

you're gonna have the ko demands the

demands that we've not been able to meet

as well as the number of the month the

total ones and then

the okay demands the amount that we've

been able to meet

we can see this information in both of

these graphs first of all a pie chart

we have uh both the okay and the ko ones

and we can see how many we've able to

meet compared to how how many we've not

been able to meet

and then we can check this data per

period

but we can see in blue our total demands

and then yellow the ones that we've

been able to meet and in green the ones

that we've been able to meet

so we've used this data that has a few

demands in period one

we had to consider that it takes one

period to order another period to

deliver

so we purposely added demands on period

one

uh that we were not able to meet because

we didn't have uh

initial stock levels we didn't define

this this is on in purpose

so we can show how and not meeting

demands would look like on our demo

however our three set of warehouses is

being is capable enough

to see uh to handle all the rest of the

demands

for the next different periods

so in the second stage of the of the

demo uh working with the supply chain

we've seen

uh the input the last input view for the

expected demand the actual real demand

for our problem

we've been able to see uh the delivery

data

uh over a different a table with the

different deliveries for this expected

demand

as well as we were able to see warehouse

per warehouse

the different stock levels for the

product units

again we were able to see this data on a

map view

we were able to easily contextualize

our geographical distribution and

finally we're able to have a better look

at how good a solution was using our kpi

analysis view

with the kps that we've defined again

this is a case to case basis where

each different customer have different

needs uh therefore different kpis

finally we're going to go into our last

stage of the b of the demo we're going

to go into the results analysis view

so we were able to see a complete

analysis

of an execution however if we

go to for instance foreign simulations

we could try and make different forecas

make different solutions modifying our

forecast

as an example of things we could do

to introduce this a little bit but what

would happen if our forecast was higher

uh how we were able to handle that with

our current data set

as well as we could go to the warehouse

configuration

and we could modify um

our availability or a different costs

that we had on our product

uh if we try to contextualize how this

would be useful

on a real real scenario situation

i want to go back to something that

we've seen before on the outputs

on the map view uh in here

we could see that from our three

different warehouses

warehouse one was not used compared to

warehouse 2 was used a little bit

and where house 3 was heavily used to

cover all of our different supplies

uh or to meter all of our demands

so let's consider what would happen if

our warehouse 3 the one in valencia was

not

able to cover all of the all of these

demands let's consider in a covered

scenario right that there's an outbreak

in valencia

and our workers are not being able to

work

due to copy what would happen if on

these

personal situations if we were to close

temporary warehouse three so we're gonna

go back into the warehouse configuration

and we would run again the data marking

this warehouse 3 as unavailable

uh before we done so and we're going to

explain how to

do this easily

give it more value to the application

you're gonna go back to the scenario

manager

okay we're gonna remove both the company

plan first stage

and the second stage this one's where

like uh

midpoints of the company plan we're

gonna still stick with the complete one

that one that has like the solution seen

to the full list

and we have here this alternate plan so

we create a new scenario

we've made a copy of the previous one

where we've modified this data

so we're going to add this uh

alternative plan that we see up here

right and we're going to put this

alternate plan as the first one so we

can see the data on the views

so if you go to this alternate plan

right we've considered this scenario

where we are not able to use warehouse

three okay and then uh we've gone into

demand

the remain view and we've also clicked

on complete run

to execute the model from start to

finish

uh if you take a look a little bit at

the map view

now you can see that warehouse three is

not being used

in the prolonged set we're gonna we're

using the other different warehouses

in this case warehouse two to supply

most of our

uh customers um the reason why

uh we're using we're not uh we were

heavily

relying on warehouse three and our heavy

rain warehouse two

is mostly due to a cost but we can

easily understand this

if we go to our last view in the

analysis tab in the scenarios view so

if you click over here on the scenarios

b what we're gonna see

here for every single scenario that i

have added to myself up here which i

have the alternative plan

and the original one that we saw before

i'm gonna do a comparison

of the different kpis that i've

evaluated before

for each different scenario it's a very

powerful tool that's gonna allow me to

compare

the first actual real scenario with

these hypothetical scenarios

so we have in here we have first a cost

table

we have the different cost as rows and

we have the different cost

per scenario as our columns we have

first this column for the

alternative plan and this one for the

original company plan

we can see this information in this

graph easily

and we can see that not using warehouse

3 has increased the cost

for our order cost our shipment cost

the stocking cost looks remains the same

however the total cost in general has

increased

not using the warehouse 3 it's going to

make the solution be

higher cost because warehouse 3 was

less expensive to use compared to the

other ones

uh we can also as well see the second

kpi but we could compare the service

level agreement

for the different plans however for our

current dataset we're able to

supply all of our customers with both

warehouses

as well as with the three warehouses

so in this last stage of the demo on

this results analysis

we've added this hypothetical scenarios

we've shown

how we can create these new hypothetical

scenarios

and use the tool to its fullest or we

could easily

modify very quick

i consider different real life

situations

again at the end we've seen this

scenario comparison

that it uh that allows us to make these

quick decisions and comparison

between these scenarios uh where each

company

may use this potent tool to their

fullest to

analyze quick that would happen in

different scenarios

so as a summary of what we've explored

today

uh well first of all we're going to

start with the necessity

we needed flexibility and stock levels

to be able to handle this uncertainty in

the demand

after we had the stock levels we were

worried about our

day-to-day supply chain management it

was very important

to have efficient stock management in

our warehouses

to meet these problems necessities our

proposal

was an application developed with fico

express insight

so we obtain out of this application is

a user friendly

and intuitive interface that allows us

to interact with the pro

with the problem and the solution uh

and which obtain a clear view of your

subjectives

uh and enhancements to the problem

solutions we were able to

easily tell and understand all the

different

um aspects of the problem we were able

to understand

and have an in-depth look into all of

them

into these objectives as well as we also

had the visual analysis

of any selective scenarios that we chose

uh through these kpis that we've defined

and

and that is all for this uh presentation

uh thank you very much for attending or

listening to us again we have some

contact information here if you wanna

address us

uh any quick question we wanna follow up

or anything we'd be very happy to

to answer that and well thank you all

for watching