| Summary: | 碩士 === 輔仁大學 === 應用統計學研究所 === 98 === The concept of “Prevention is better than cure” is entrenched. With the increasing income, the health foods become popular. The health foods market is an increasingly targeted by food marketers. There are many different types of health foods, one of the most common types is the health tablet.
In order to examine stable factors that influence willingness to buy health tablets. This study used resampling technique to split up into 100 dataset. This study took advantage of MARS method, traditional statistical method, and MARS merge traditional statistical method to find important factors which influence willingness to buy health tablets. Finally, this study used selected factors from above methods to construct the logistic regression prediction models. The result shows that MARS merge traditional statistical method has better performance than MARS, traditional statistical method in prediction ability.
The results of logistic regression prediction model are shown as follow: Intensity of environmental awareness, smoking, cosmetology spending, married- age (integrate factor), amount of personal assets, amount of drinks bought recent year, oral drugs- topical drugs (integrate factor), and county are significant predictor for willingness to buy health tablets. The accuracy rate is 71.76%. This model finds target customers could improve the gain of twice and non-target customers could add 10% accuracy rate. This model could help the industries to know the characteristics of customers who would buy health tablets. The results would also help the industries to make marketing strategies, decrease costs, improve promotion effectiveness and create greater profitability.
|
|---|
