Intelligent Sale Prediction System

MASTER THESIS COMOUTER ENGINEERING 2009

ISPS

Table OF CONTENTS

Abstraction

List of Figures

List of Tables

Recognition

1: Introduction

1.1 Aim and Objective

1.2 Scope of the Undertaking

2: Literature Reappraisal

2.1 Benefits of Sale Forecasting

2.2 Economicss Laws Affecting Product Sale

2.2.1 Price of the Merchandise

2.2.2 The Consumer ‘s Income

2.2.3 Consumer Awareness

2.3 Fuzzy Logic

3: The Approach

3.1: Fuzzification

3.1.1 Input 1 Purchasing Power

3.1.2 Input 2: Awareness:

3.1.3 Input 3: Monetary value

3.1.4 End product: Sale

3.2 Rules Evaluation

3.3 Collection

3.4 Deffuzification

4: Results & A ; Discussion:

4.1 ISPS Consequences

4.2 Comparison ISPS Sale Results with Actual Sale

5: Mentions

List of Figure

List of Table

1: – INRODUCTION

Intelligent system that could foretell sale of bing merchandise of a company. It will be a job based, research oriented thesis chiefly concentrating country of concern selling, gross revenues direction and merchandise publicity. ISPS will be capable to calculate gross revenues volume that is considered as the most effectual tool to capture maximal market portion. In the field of concern direction intelligent gross revenues prediction is consider as a smart manner to falter upon rivals. Our theoretical account will move as a ‘Intelligent determination maker’ to put gross revenues marks.
Prediction of sale is critical and complicated procedure which includes many obscure parametric quantities, fuzzy adept system to get by with this challenge. Fuzzy Logic is a better beginning to supply methodological analysis to work with uncertainness and imitate worlds concluding, the gross revenues factors are introduced. Then a fuzzy expert system is designed to calculate sale of the bing merchandise sing the related parametric quantities.
To formalise ISPS I have to work on following sectors:
Study of research work in the similar field
Fuzzy systems
Fuzzy logic rank maps
Fuzzy regulations and Fuzzy illation
Working on different Methodologies
Multifaceted character of Sale
Modeling the job with fuzzed logic
Input signals Variables:
1. Customer Purchasing power
2. Awareness












3. Monetary value
End product:
Gross saless volume



1.1 AIM AND OBJECTIVE

The purpose of this thesis is to construct a fuzzy expert system for foretelling sale volume on the footing of assorted factors impacting the sale volume. Gross saless calculating is gauging what a company ‘s future gross revenues are likely to be based on Buying power, Awareness and monetary value. Gross saless prediction is a self-assessment tool for a company. You have to maintain taking the pulsation of your company to cognize how healthy it is. A gross revenues prognosis studies, graphs and analyzes the pulsation of your concern. It can do the difference between merely lasting and being extremely successful in concern. It is a critical basis of a company ‘s budget. The future way of the company may rest on the truth of your gross revenues prediction.

1.2 SCOPE OF THE PROJECT

The undertaking is concentrating on following issues.
Design a fuzzy expert system ( ISPS ) for sale volume prognostication. The combination of these benefits may ensue in:
§ Increased gross
§ Increased client keeping
§ Decreased costs
§ Increased efficiency




2: Literature Reappraisal

ISPS is gauging gross revenues what a company gross revenues will be in future on different factors. In this ISPS we will see merely three factors Advertisement, Buying power and monetary value. The importance of sale prediction for company is in hopes of placing forms so that gross and hard currency flow can be increased.

Basically, gross revenues prediction is analysing all parts of a concern from entire stock list to the strengths and failings of sales representative. It is really good for directors to take determination. They must be spirited when measuring the competition and how they can excel the competition to better run into the demands of the mark market.

Intelligent Gross saless anticipation system ( ISPS ) is a self-assessment tool for a company. You have to maintain taking the pulsation of your company to cognize how healthy it is. ISPS is really helpful for lasting and being extremely successful in concern. It is a critical basis of a company ‘s budget. Gross saless anticipation system is helpful for company to rearrange the gross revenues marks for approaching months and future way of the company with truth.

2.1 Benefits of Sale Forecasting

• Enhanced hard currency flow
• Knowing when and how much to purchase
• In-depth cognition of clients and the merchandises they order
• The ability to program for production and capacity
• Ability to find the expected return on investing



The combination of these benefits may ensue in:

• Increased gross
• Increased client keeping
• Decreased costs
• Increased efficiency


In the field of concern direction intelligent gross revenues prediction is consider as a smart manner to falter upon rivals. Our theoretical account will move as a ‘Intelligent determination maker’ to put gross revenues marks. In the ISPS fuzzy expert system will state steer us in puting the mark of the company. This ISPS will establish on merely three factors ( Price, Awareness, and Purchasing Power ) . These three factors are most of import in puting the gross revenues mark.

2.2: Economicss Laws Affecting Product Sale

Even though the focal point in economic science is on the correlativity among factors ( Buying Power, Awareness and Price ) which are impacting the merchandises demand and how much consumers are willing and able to purchase, it is of import to analyze all of the factors that affect the sale for a good.
These factors include:

2.2.1 Price of the Merchandise

There is an opposite ( negative ) relationship between the monetary value of a merchandise and the sum of that merchandise consumers are willing and able to purchase. Consumers want to purchase more of a merchandise at a low monetary value and less of a merchandise at a high monetary value. This opposite relationship between monetary value and the sum consumers are willing and able to purchase is frequently referred to as The Law of Demand.

2.2.2 The Consumer ‘s Income

The consequence that income has on the sum of a merchandise that consumers are willing and able to purchase depends on the type of good we ‘re speaking about. For most goods, there is a positive ( direct ) relationship between a consumer ‘s income and the sum of the good that one is willing and able to purchase. In other words, for these goods when income rises the demand for the merchandise will increase ; when income falls, the demand for the merchandise will diminish. We call these types of goods normal goods.

2.2.3 Consumer Awareness

The per centum of population or mark market, who are cognizant of the being of a given trade name or company.

While doing ISPS I consider the above mentioned economic sciences affects on sale volume.

2.3 Fuzzy Logic

Fuzzy logic is a signifier of multivalent logic derived from fuzzed set theory to cover with concluding that is approximative instead than precise. In contrast with binary sets holding binary logic, besides known as chip logic, the fuzzed logic variables may hold a rank value of non merely 0 or 1. Merely as in fuzzy set theory with fuzzed logic the set rank values can run ( inclusively ) between 0 and 1, in fuzzed logic the grade of truth of a statement can run between 0 and 1 and is non constrained to the two truth values { true ( 1 ) , false ( 0 ) } as in authoritative propositional logic. [ 1 ] And when lingual variables are used, these grades may be managed by specific maps, as discussed below.

The term “ fuzzed logic ” emerged as a effect of the development of the theory of fuzzed sets by Lotfi Zadeh [ 2 ] .

In 1965 Lotfi Zadeh proposed fuzzed set theory [ 3 ] , and subsequently established fuzzy logic based on fuzzed sets. Fuzzy logic has been applied to diverse Fieldss, from control theory to unreal intelligence, yet still remains controversial among most statisticians, who prefer Bayesian logic, [ commendation needed ] and some control applied scientists, who prefer traditional two-valued logic. [

What Are Fuzzy Inference Systems?

Fuzzy illation is the procedure of explicating the function from a given input to an end product utilizing fuzzed logic. The function so provides a footing from which determinations can be made, or forms discerned..

Fuzzy illation systems have been successfully applied in Fieldss such as automatic control, informations categorization, determination analysis, adept systems, and computing machine vision. Because of its multidisciplinary nature, fuzzed illation systems are associated with a figure of names, such as fuzzy-rule-based systems, fuzzed expert systems, fuzzed mold, fuzzed associatory memory, fuzzed logic accountants, and merely ( and equivocally ) fuzzed systems.

Mamdani ‘s fuzzed illation method is the most commonly seen fuzzed methodological analysis. Mamdani ‘s method was among the first control systems built utilizing fuzzed set theory.

Mamdani-type illation, as defined for the tool chest, expects the end product rank maps to be fuzzed sets. After the collection procedure, there is a fuzzed set for each end product variable that needs defuzzification. It is possible, and in many instances much more efficient, to utilize a individual spike as the end product rank map instead than a distributed fuzzy set. This type of end product is sometimes known as a singleton end product rank map, and it can be thought of as a pre-defuzzified fuzzy set. It enhances the efficiency of the defuzzification procedure because it greatly simplifies the calculation required by the more general Mamdani method, which finds the centroid of a planar map. Rather than incorporating across the planar map to happen the centroid, you use the leaden norm of a few information points. Sugeno-type systems support this type of theoretical account. In general, Sugeno-type systems can be used to pattern any illation system in which the end product rank maps are either additive or changeless.

Multiple Arrested development

THE Approach

The ISPS is based on four stairss.

In the ISPS I formulate the function of three inputs to an end product. The procedure of ISPS invovles rank maps, Logical Operation and if- so regulations. Mamdani fuzzed illation system is used to foretell sale volume.

  1. Fuzzification
  2. Rule Evaluation
  3. Collection
  4. Deffuzification

The flow chart of the ISPS is given below in figure …………… .

Input 1

PurchasingPower

Rule1

Rule 2

End product

Sale

Input 2

Awareness

Sum

Rule 3

Input 3

Monetary value

Rule 27

Figure: 1.1:

3.1: Measure 1: FUZZIFICATION

Under FUZZIFICATION, the rank maps defined on the input variables are applied to their existent values, to find the grade of truth for each regulation premiss.

In ISPS I use three inputs Buying Power, Awareness and Price. On these three of import factors the ISPS foretell about sale.

The generalized bell map is used it depends on three parametric quantities a, B, and degree Celsius as given by

[ image ]

3.1.1 Input 1 Purchasing Power

Buying power is the figure of goods/services that can be purchased with a unit of currency.

Fuzzy Sets: Low Purchasing power, Medium buying power, High buying power.

Here you can see the scopes of lingual variables and their rank values.

Low 2000 to 2400

Master of education 2300 to 2700

3.1.2 Input 2: Awareness:

The acquaintance of a merchandise, trade name name, company, new construct or hallmark.

Fuzzy sets: Low Awareness, Medium Awareness, High Awareness.

3.1.3 Input 3: Monetary value

The sum of money needed to buy something

Fuzzy Sets: Low Price, Medium Price, High Price

3.1.4 End product: Sale

Fuzzy sets: really low sale volume, low sale volume, Medium sale volume, High sale volume and Very High sale volume.

3.2 STEP 2: RULES Evaluation:

In ISPS I make 20 seven regulations. In the first nine regulations when buying power of peoples is high so they are non concious about monetary value.

Rules

Graphic View of Rules

In the Graphical position of the regulations shows the sale of the whole fuzzy expert sytem. These 20 seven regulations are based on fuzzed illation diagrams, which I show above.

The variables buying power, Awareness, Price and sale volume are shown on the top of the graph. In the lower left we enter the values of our input variable, In ISPS I use three input variables so we enter the values as [ 2600 50 50 ] . 2600 is the Buying power, 50 is Awareness and the last 50 is the monetary value of the merchandise. We can alter the values of our input variables by snaping on the secret plans ( Red Line ) for each input. The ruddy line will be move horizontally.

Where the index line stand foring service crosses the rank map line “ service is hapless ” in the upper-left secret plan determines the grade to which govern one is activated.

A xanthous spot of colour under the existent rank map curve is used to do the fuzzy rank value visually evident.

Each of the word pictures of each of the variables is specified with regard to the input index line in this mode. If you follow govern 1 across the top of the diagram, you can see the consequent “ tip is inexpensive ” has been truncated to precisely the same grade as the ( composite ) antecedent—this is the deduction procedure in action. The collection occurs down the 3rd column, and the attendant sum secret plan is shown in the individual secret plan looking in the lower right corner of the secret plan field. The defuzzified end product value is shown by the thick line go throughing through the aggregative fuzzy set.

You can switch the secret plans utilizing left, right, down, and up. The bill of fare points allow you to salvage, open, or redact a fuzzy system utilizing any of the five basic GUI tools.

The Rule Viewer allows you to construe the full fuzzy illation procedure at one time. The Rule Viewer besides shows how the form of certain rank maps influences the overall consequence. Because it plots every portion of every regulation, it can go unmanageable for peculiarly big systems, but, for a comparatively little figure of inputs and end products, it performs good ( depending on how much screen infinite you devote to it ) with up to 30 regulations and every bit many as 6 or 7 variables.

The Rule Viewer shows one computation at a clip and in great item. In this sense, it presents a kind of micro position of the fuzzed illation system. If you want to see the full end product surface of your system—the full span of the end product set based on the full span of the input set— you need to open up the Surface Viewer. This spectator is the last of the five basic Fuzzy Logic Toolbox GUI tools. To open the Surface Viewer, choice Surface from the View bill of fare.

Each of these input lingual variables have three fuzzy values Low, Medium and High and the end product ( Sale ) fuzzy values are really low, Low, Med, High and really High. The historical informations are used to find the sphere of all rank maps.

In the regulation base 27 IF-THEN regulations are used for all possible combinations of the fuzzed values of the lingiustic variables.

While doing regulations I consider the economic sciences Torahs.

3.3 STEP 3: Collection

3.4 STEP 4: DEFUZZIFICATION

Concluding measure is the DEFUZZIFICATION, which is used when it is utile to change over the fuzzed end product set to a sharp figure. There are many defuzzification methods that we can utilize. Two of the more common techniques are the CENTROID and MAXIMUM methods. In the CENTROID method, the chip value of the end product variable is computed by happening the variable value of the Centre of gravitation of the rank map for the fuzzed value. In the MAXIMUM method, one of the variable values at which the fuzzy subset has its maximal truth value is chosen as the chip value for the end product variable.

To foretell the sale value centroid defuzzification expression is used. Consequently, the predicted end product, sale, is computed as

4: Consequence:

4.1 ISPS Consequences

Month 2008

ISPS Sale Results

Jan

435000

Feb

445000

Mar

450000

Apr

476000

May

481000

June

484000

Jully

494000

Aug

483000

Sep

492000

Oct

495000

Nov

510000

Dec

512000

4.2 Comparison ISPS Sale with Actual Sale

Month 2008

Actual Sale

Predicted Sale

Jan

448000

435000

Feb

454000

445000

Mar

459000

450000

Apr

488000

476000

May

497000

481000

June

497000

484000

Jully

499000

494000

Aug

480000

483000

Sep

489000

492000

Oct

503000

495000

Nov

526000

510000

Dec

528000

512000

Comparison Shown in Graph

On the X-axis shwon the months from January to December and on the Y-axis demoing the sale volume scope from 100000 to 600000 on the above reference graph. Here you can see Fuzzy Expert System successfully predict the gross revenues. The predicted sale consequences are really near to the existent sale volume.

Comparison WITH MULTIPLE REGRESSION

Adept systems have been used to work out a broad scope of jobs in spheres such as medical specialty, mathematics, technology, geology, computing machine scientific discipline, concern, jurisprudence, defence and instruction. Within each sphere, they have been used to work out jobs of different types. Types of job involve diagnosing ( e.g. , of a system mistake, disease or pupil mistake ) ; design ( of a computing machine systems, hotel etc ) ; and reading ( of, for illustration, geological information ) . The appropriate job work outing technique tends to depend more on the job type than on the sphere.

PDAMum as an expert system might do errors, but it is less than a human did. Furthermore it ever performs systematically, ne’er become tired or bored. Other clear different is PDAMum can be use anywhere anytime compared to human.

User may clear up their PDA synchronism job with immediate response and recover such dependable diagnosing through PDAMum. This characteristic will help them to acknowledge the causes that disallow their PDA synchronism. They may inquire PDAMum why they being inquire such inquiry during the interaction procedure.

In future, PDAMum can be improve by convert it as web base application. Alternatively of working base entirely, PDAMum can be reach through web site which wider the entree location. User from assorted PDA proprietors may entirely utilize individual expert system to get the better of their PDA’s job.

PDAMum besides should hold better account installation. This can be achieve if it has more regulations to trades with other PDA job such as sound and power direction. With extra cognition, PDAMum can work out and polish the job deeper, so that user may see other possibility that blocked their Personal digital assistant from perform good.

Other rational enlargements are image spectator and uncertainness factor. PDAMum can be utilize better if it can see meaningful image that will back up it’s name and determination. Uncertainty besides important since user sometime can non show their feels. With Certainty Factor ( CF ) capableness, PDAMum will increase user assurance and convert them to do right pick.

Decision

PDAMum like other expert system will travel a enormous stages from simple expert system to the complex multipurpose systems. Hybrid expert system and together with fuzzy expert system can be seen as a new techniques that be used by research workers recently. Execution of expert system in such Fieldss is greatly influenced by techniques and methods from adaptative hypertext and interactive multimedia. Features of personalization, user mold and ability of adaptative towards environment will go great challenges to settle. It can be used as a guideline to advance an expert system in assorted maps.

In future, PDAMum can be used together with unreal nervous webs, fuzzed logic, familial algorithms and other methods of Artificial Intelligence. These methods allow taking into account their advantages in the intentional system and, hence, new designed systems are more powerful instruments to ease assorted undertakings that require instant, accurate and dependable consequences

hypertext transfer protocol: //www.econport.org/content/handbook/Demand/Factors.html

hypertext transfer protocol: //en.wikipedia.org/wiki/Fuzzy_logic

hypertext transfer protocol: //www.mathworks.com/access/helpdesk/help/toolbox/fuzzy/index.html? /access/helpdesk/help/toolbox/fuzzy/fp351dup8.html & A ; hypertext transfer protocol: //www.google.com/search? hl=en & A ; rlz=1R2SNYK_en-GBSE330 & A ; q=mamdani+fuzzy+logic+steps+detail & A ; aq=f & A ; oq= & A ; aqi=