Bài giảng tối ưu hóa Bài 1 Giới thiệu Ph.D Lê Minh TâmHUFI

What is “Running experiment” RE ? RE: to understand a system by disturbing its factors  RE efficiently: doing the least amount of works and getting the most amount of information  RE

Design of experiments

Tâm Minh Lê, Ph.D Faculty of Food Technology, HUFI

1.01 Introduction

When was the last time you ran

What is “Running experiment” ?

 We do it all the time

What is “Running experiment” ?

 We do it all the time

What is “Running experiment” (RE) ?

 RE: to understand a system by disturbing its factors

 RE efficiently: doing the least amount of works and getting the most

amount of information

 RE at work (R&D staff): money and time consuming, we have to run

experiment correctly at the first time

Running experiments in industrial contexts

 Boiling water: pick the setting that uses the lowest total energy

 Heating setting: low vs high

 Type of pot: metal vs glass

 Making use of lid: with vs without

Running experiments in industrial contexts

 Planting mushroom

 Light intensity: low vs high

 Watering/Humidity: low vs high

 Type of soil: A vs B

 Fertilize: C vs D

 etc

Summary: why do we have to run experiments ?

 To improve an objective

 To improve an objective, we must understand its system

 To understand the system, we might adjust multiple factors (a lot of

work)

 Running experiment allows us to do the least amount of works and

getting the most amount of information

 How about your context ?

1.02 Terminology

Outcome = Response

 What we measure after the experiment is finished

 It is the aspect we want to improve

Factors = Variables

 Things influencing the outcome that you want to change

 Typically we change many factors (2, 3, etc.)

 Types of factors:

 Quantitative/Numeric: toC, mL, …

 Qualitative/Categorical: color, with or without an ingredient

Example

 What are the outcome(s), factors and objective in this experiment ?

 Various outcomes are possible in our experiment

 height of mushroom

 number of mushroom (i.e., productivity, yield)

 quality (e.g., freshness, structure)

 Most of the outcomes are quantitative

 Qualitative outcomes are possible: color, budding … (but infrequently)

 Various factors are possible to adjust:

 Light intensity: low vs high

 Watering/Humidity: low vs high

 Type of soil: A vs B

 Fertilize: C vs D

One of the following objectives:

 maximize the profit ()

 maximize the quality of mushroom ()

 minimize the energy to produce a product ()

Another example

 Using butter produced by Company A or Company B ?

 Outcomes: the taste

 Objective: the taste of two recipes are the same

 Remark: Maximizing the profit (because the price of the « butter »

produced by Company A is too expensive) is the objective of the

company This is not the objective expected from this experiment

 Factors: changing processing as toC, time, …

 What are the objective, outcomes and factors of your own experiments ?

Summary: Do’s and Don’ts

An experiment

 might has any outcomes Do measure as many outcomes as possible

 needs at least one factor to adjust Don’t fixate on changing One Factor

At a Time (OFAT)

1.03 Interpreting results

Context – Profit of a selling store

 Whether the profit in every Monday (of the store) is effected if we

change the dimmer and the price of the product

Context – Profit of a selling store

What are the outcomes, objective and factors of this experiment?

Context – Profit of a selling store

 Outcome:

 profit made in one day when selling a specific product

 quantitative outcome, because profit = total income – total expenses

 Objective: maximize the profit ()

 Factors:

Low High

Dimmer

Low High

Price

How will dimmer affect the outcome :

 Does a higher light increase the profit ?

 Does a higher light decrease the profit ?

 Does a higher light actually effect the profit ?

How will price affect the outcome ?

 Does a higher price increase the profit ?

 Does a higher price decrease the profit ?

Collecting data 6 ‒ Light: high

Here are the results

Summaryzing the data

Dimmer (%) Price Profit ($)

Analyzing the data

Analyzing the data

Analyzing the data

Analyzing the data

Analyzing the data

Analyzing the data

 What if one day we have a bad weather ?

 What if we re-run the experiment, will we have the same results ?

1.04 DICUSSION ON OFAT

Context – Profit of a selling store

 Outcome: profit of every Monday in a selling store

 Objective: maximize the profit ()

 Factors

 Factor A – the amount of light (dimmer): 50% or 75%

 Factor B – the price: 7.79 c or 8.49 c

 At keeping Dimmer factor at high level:

Price effect (Effect of factor B)

 At keeping Dimmer factor at low level:

370 − 490 = −120 $

Think about your own experiments

 pop corn

 pan cake

 black garlic

 etc

Group working

 Summarize your today lesson and

 Do the following exercises

Exercise 1

Exercise 2

Exercise 3

To investigate this Rishat has

considered the following two factors:

 S = Quantity of sugar in vase

water, low level was 0 g/ml, and

high level was 3 g/ml

 V = Quantity of vodka in vase

water: low level was 0 g/ml, and

high level was 3 g/ml