透過您的圖書館登入
IP:3.137.161.222
  • 學位論文

以類神經網路模組探討人格特質、工作特性、人際關係及組織氣候對其工作滿足之影響

Research on effects of job satisfaction with predictors of personality、job characteristics、interpersonal relationship and organizational climate by using neural network module

指導教授 : 洪英正 黃曼琴

摘要


類神經網路能模擬人類思考模式,其優點除了能解決非線性之問題外,它還具有學習與分辨複雜資料之能力。因此,本研究希望藉由類神經網路之上述特性來預估人類之行為模式,並建構出一套評估適任人選之模組。 本研究彙集國內外學者研究,搜集整理其中影響工作滿足較為顯著的重要因子,例如:人口統計變項、人格特質、工作特性、人際關係和組織氣候等,將影響因子輸入類神經網網路中加以訓練,並建立起類神經預測模組。本次問卷之研究對象則以台灣地區之工作者為主,問卷係採用便利抽樣方式,此次問卷共發放429份,有效回收率為67.6%。經本研究分析後,可彙集底下幾點結論: 一、事先透過統計手法做篩選動作,並不會對於類神經網路之辨識率有顯著提升作用; 而類神經網路與迴歸方程式在工作滿足上之辨識率,也無顯著差異。 二、不管使用類神經網路,或是迴歸統計之分析手法,在內部滿足之二段分類程度時,兩種模組之預測能力均達九成以上的水準。 三、研究中發現,類神經網路的自我辨識率超過94.88%,代表類神經網路應用於工作滿足辨識模組時,其自我學習能力很高。 本研究比較類神經網路模組與迴歸方程式之辨識率後發現,類神經網路模組在自我樣本的辨識率非常高,所以利用類神經網路模組來辨識已知樣本,效果是相當顯著,甚至比迴歸方程式還要高;但類神經網路模組在辨識未知樣本上,其辨識率就比迴歸方程式來得差,顯示類神經網路模組在學習新樣本的過程中,還是有其改善的空間。

並列摘要


Neural networks can simulate the thinking module of the mankind, and the advantages are able to not only solve the non-linear problem but also study and distinguish the complicated materials. For this reason, this research hopes to evaluate the human behavior in advance with the above-mentioned characteristics of neural networks, and to construct the module. The research gathered the literature from domestic and international scholar, and collected the important factor affecting job satisfaction apparently. For example, demographics, personality, job characteristics, interpersonal relationships and organization climate all will influence employees’ job satisfaction, and we put them as input variables in neural network for training as a predictive module. The questionnaire of this research selects workers in Taiwan as samples. We collected 429 copies of questionnaire, and the effective return rate is 67.6%. After analyzing all questionnaires, there are some of following conclusions: 1. By using the statistic technique in advance, it will not have remarkable function for the discriminative rate of neural network. By the way, there is no difference in the discriminative rate of job satisfaction between the neural network and regression model. 2. No matter using neural network module or regression model at two degrees of internal job satisfaction, the correct rate of two kinds of methods are all more than 90%. 3. We find the self discriminative rate of neural network exceed 94.88% which representing the self-learning ability is very high when applies to the job satisfaction. According to the research, we find the discriminative rate of neural network in training sample is very high after comparing the discriminative rate between neural network and regression model. For this reason, we use neural network to discriminate the known samples, and the result is quite apparent, even higher than regression model. However, neural network module in discriminate unknown samples is worse than the regression model. It means that there are still some improved spaces for neural network module to improve when apply it at unknown samples.

參考文獻


25. 張子仁 (2002)。模糊類神經網路影像校準方法應用在精密樣板比對。中原大學機械工程研究所碩士論文。
24. 植懷信 (2005)。台北地區檔案管理人員工作滿意度之研究。台灣大學國家發展研究所碩士論文。
17. Su-Chao Chang and Ming-Shing Lee. “Relationships among Personality Traits, Job Characteristics, Job Satisfaction and Organizational Commitment - An Empirical Study in Taiwan” The Business Review, Cambridge. Hollywood: Dec 2006,6, 201-207
17. 陳明崇 (2002)。國民中小學專任人事人員工作特性與工作滿意之研究--以台灣北部五縣市為例。國立台北師範學院國民教育研究所碩士論文。
15. 陳文忠 (2003)。組織氣候、學習動機與學習行為對人力資源績效影響之研究-以國內企業員工回流教育為例。國立成功大學高階管理碩士在職專班碩士論文。

延伸閱讀