pression of customer dissatisfaction, posted on a review website where customers who actually purchased personal computer related products post their reviews as shown in Table 2.

The study, then, classifies the posted customer complaints based on the six quality characteristics defined in the ISO/IEC9126-1, obtains from questionnaires a weight for each quality characteristic to represent the importance of quality characteristic for the customer, and applies the weights to the six quality characteristics, and quantifies the degree of customer satisfaction for each quality characteristic as shown in Table 3.

This study calculates the observed total customer satisfaction by using customer satisfaction of each six quality characteristics based on the technique that I showed for the beginning in a precedent study [11].

This study second collects price of each product as as-signed property and inherent attributes of target products, i.e., CPU, HDD capacity, weight of body, Drive time, as shown in Table 1.

The study performs correlation analysis among of the degree of price and inherent attributes of LPCs, and confirms that the attributes of them are mutual independent or not.

In addition, the study develops a model that actually predicts the degree of total customer satisfaction of the

target products from the degree of price and inherent attributes of product, and the study performs multiple regression analyses.

For performing a multiple regression analysis, the degree of observed total customer satisfaction of product has assigned to the objective variable, and price and inherent attributes of product has assigned to the explanatory variables.

The possibility of whether or not the degree of observed total customer satisfaction of products can be derived from the degree of price and inherent attributes of target product during the system design phase was verified.

Finally, the study discusses the validity of estimated prediction models based on the significance of the developed prediction model and possibility of application of proposed prediction model and approaches.

3.1. Target Data

In recent years, due to the explosion of the Internet, purchasing behaviours of customers have significantly changed. For example, an increasing number of customers order a product directly from an electric commerce site without visiting brick-and-motor shops while remaining at home. The degree of customer satisfaction is a measure use din marketing that represents how a product or service produced by a company meets or surpasses customer expectation. This study focuses on online reviews posted on the Internet, an effective alternative to face-toface interviews of customers, and uses the online negative reviews of a system product as data to investigate. This study collects and uses online reviews of products posted at a web-site, kakaku.com [13] as customer’s expression of his/her dissatisfaction of system products. Table 2 shows the part of collection data concerning negative review from web-site, which total number of type of LPCs is 35 and total number of review is 457.

From Table 2, this study counts the number of online negative reviews for each concrete category of interest of LPCs from the view point of the six quality characteristics. This study collects and classifies online negative reviews from the view point of the six quality characteristics in this manner. Furthermore, for each product, this study obtains the degree of importance for each of the six quality characteristics taking into account the interest of attribute of LPCs (i.e., weight for) correspond to the six quality characteristic as shown in Table 3. Table 3 shows the questionnaires and example of result of importance of quality needs for each six quality characteristics obtain from questionnaires.

The questionnaires asked customers the question of “in purchasing a LPC, what attributes are important?” and have the customers assign a numeric order number between 1 and 15 based on the importance;

The order number is 1 for the most important for the quality characteristic, and the weight for the six quality characteristics are normalized in the range from 0 to 1.

This study quantitatively calculates the importance of customer needs for each quality characteristics.

For example, the importance of functionality as M is obtained from the following Equation (1).

(1)

M: Importance ratio of Functionality (weight)

aaa1_n: Order number of importance (aaa1 = 1 - ON)

aaa2_n: Order number of importance (aaa2 = 1 - ON)

n: Number of customers (n = 1 - TN)

ON: Maximum order number (ON = 15)

TN: Total number of customers (TN = 61)

From Table 3, TN is the total number of customers, which total number of customers is 61, and p is the weight for the efficiency determined from the questionnaires.

This study also focuses on laptop computers (LPCs). Reasons for choosing LPCs are, at first, LPCs have attribute and characteristics that correspond to the six quality characteristics, and there is a large amount of data available on the non-functionality and non-quantitatively requirements on online review web-sites.

Table 1 shows the part of collection data concerning inherent attributes of LPCs from web-site, which total number of type of LPCs is 35. Inherent attribute refers to the degree to which attributes of target product have the intrinsic potential to satisfy stated and implied needs when LPCs is used under specified conditions. From the inherent point of view, Attribute of LPCs is product itself, in particular to shown in Table 1.

Above consideration, price is not inherent attribute but assigned attribute. If we wish to evaluate the customer satisfaction from the view point of product quality, we should use only inherent attribute of target product itself. Then, this study use only inherent attributes of target product without price.

3.2. Degree of Customer Satisfaction

Table 4 shows the parts of result of degree of the ob served customer satisfaction for each quality characteristics.

By applying the weight for each of the six quality characteristics, this study quantitatively calculates the degree of customer satisfaction for each six quality characteristics. For example, the degree of customer satisfaction for efficiency as DD is obtained as following Equations (2) and (3).

(2)

(3)

This study quantitatively calculates the degree of observed total customer satisfaction from the view point of

six quality characteristic included in the ISO/IEC9126 quality model.

The degree of observed total customer satisfaction is obtained based on the consideration of independency among six quality characteristics as following Equations (4) as shown in Table 4.

(4)

TS: Observed total customer satisfaction

This study also predicts the degree of total customer satisfaction and price of product. The degree of predicting total customer satisfaction sis obtained by using inherent attributes of LPCs as following Equations (5). And, the degree of price of product by using inherent attributes of LPCs as following Equations (6).

(5)

PS: Predicting total customer satisfaction

r_n: partial regression coefficient (n = 1 - 14)

(6)

PP: Predicting price of product

r_n: partial regression coefficient (n = 1 - 14)

4. Verification

4.1. Correlation Matrix between Observed Total Customer Satisfaction and Attributes

Table 5 shows the result of correlation analysis among inherent attributes and observed “Price and total customer satisfaction” of LPCs. The correlations about “Price and some inherent attributes” are recognized. Also, the results of correlations between “total customer satisfaction and some inherent attributes” are recognized.

4.2. Multiple Regressions Analysis among Observed Total Customer Satisfaction and Attributes

Table 6 shows the three types of prediction models based on the consideration about the result of correlation analysis. From Table 6, multiple-regression analysis between an observed total customer satisfaction and the price and con cerning inherent attributes of each type of prediction model shows that the maximum value of the multipleregression coefficients and the determination coefficients are 0.5626 and 0.3165, respectively as prediction model Type 02. In addition, maximum value of F-test is 2.6853. Since it is bigger than 5% significance level F₀ = 2.5336, this study confirms that there is significance in predicting the degree of total customer satisfaction.

The cause and effect relationship between the degree of observed total customer satisfaction and those corresponding inherent attributes of LPCs could be observed.

This study verified the validity of the introduced prediction models of quantitatively predicting total customer satisfaction using the inherent attributes of target LPCs.

But, the cause and effect relationship between the de-

gree of observed total customer satisfaction and those corresponding prices of LPCs could not be observed.

Since consideration of above result, this study confirms that causal relationship are not exists among the observed total customer satisfaction and concerning price of sample products.

Table 7 shows the nine types of prediction models price based on the consideration about the result of correlation analysis shown in Table 5. From Table 7, multiple-regression analysis between the observed price of product and concerning inherent attributes of each type of prediction model shows that the maximum value of the multiple-regression coefficients and the determination coefficients are 0.7776 and 0.6046, respectively as prediction model Type 11.

In addition, maximum value of F-test is 11.4696. Since it is bigger than 1% significance level F₀ = 4.0179, this study confirms that there is significance in predicting the degree of price of LPCs.

The cause and effect relationship between the degree of observed price of product and those corresponding inherent attributes such as “d, i, k, n” of LPCs could be observed. But, the cause and effect relationship between the degrees of observed price of product and those corre sponding inherent attributes such as “a, f, h, o” of LPCs could not be observed.

Since consideration of above result, this study confirms that causal relationship are exists among the observed total price and those part of corresponding inherent attributes such as “d, i, k, n” of LPCs could be observed.

Also, the cause and effect relationships between the degree of observed “average and minimum price” of products and those corresponding inherent attributes are high.

But, the cause and effect relationships between the degrees of observed “maximum price” of products and those corresponding inherent attributes are low.

5. Concluding Remarks

Figure 4 shows the confirmed result of investigation based on the result of predicted price and total customer satisfaction. In order to understand easily, Figure 4 illustrated based on the result of multiple-regression analysis shown in Table 5. In the Figure 4, the arrow of the solid line shows that a factor and the relationship of the result were recognized from result of this analysis. On the other hand, the arrows of the dashed line have not confirmed the causal relationship based on the analysis of this study, but there is thought that there are relations from experience.

From Figure 4, it was admitted that peculiar inherent attribute of “a, f, h, o” of a product had an influence on the total customer satisfaction of the product. Furthermore, it was admitted that peculiar assigned attribute of “d, i, k, n” of a product had an influence on the price of product. On the other hand, the influence on price was not confirmed by from the inherent attribute of “a, n, f, h, o” of product.

As for this, the inherent attributes of product has the influence on the cost, and it is thought that, as a result, inherent attributes of product may influence the price as shown in Figure 4.

It is thought that the degree of total customer satisfac-

tion is not influenced by peculiar inherent attribute of “d, i, k, n”, but the price of product is influenced by the peculiar inherent attribute of product from the above mentioned result.

In other words, this study confirmed that the price of product did not influence the total customer satisfaction of the product directly.

The reason of this, information from internet includes the unsatisfactory opinion for quality of the product mainly, and, it is thought that the opinion for price is not included.

Developers may improve total customer satisfaction by concentrating the financial resources into peculiar inherent attributes of product which is cause of increasing of total customer satisfaction, and controlling customer satisfaction and it is thought that, as a result, may realize the increase of sales.

—inherent attributes of product are contributes to the total customer satisfaction of product quality.

—inherent attributes of product are contributes to a price.

—price as assigned property of product may not contribute to total customer satisfaction of product quality, but may contribute to sales directly.

Developers improve total customer satisfaction by optimize a product quality and may finally increase a sales and profit based on the limited financial resources.

Thus, it is necessary to divide an improvement of price and product quality.

And, we may increase profit by the following methods:

—optimize a peculiar inherent attribute of a product to raise customer satisfaction.

—optimize a peculiar inherent attribute of a product to decrease a cost, and realize lower a price.

—improve total customer satisfaction and realize an improvement of sales.

In the future study, the author plans to study the influence of inherent attributes and application of proposed models to each six quality characteristics.

6. Acknowledgements

The authors are grateful to members of production system research office at Graduate School of Factory of Science and Engineering HOSEI University who their contributions and support to make the discussion.

REFERENCES