在消費者資料庫中通常以RFM指標預測消費者購買行為，而其機率模型研究多以R、F所建構的機率模型結果配合顧客價值模式計算M指標之顧客終身價值，尚未有研究針對金額指標建立單獨模型以進行預測分析，因此本研究以Markov模型首先針對消費者每次消費金額做階層劃分，再根據每次交易所屬金額階層類別並跨次數交易間金額階層的轉換，與總體交易次數之比值計算交易金額階層間轉移機率（transition probability）得到Markov轉移矩陣（transition matrix），並依此作為未來顧客交易金額之預測，結果發現一般顧客朝向高金額等級轉換，或停留在較高等級的機率皆比高貢獻顧客來的大；因此若以傳統的分群方法區分高低貢獻顧客，可能無法瞭解金額單筆貢獻轉換過程中，一般顧客群也可能朝向高金額轉換的動態性，提出以動態機率模式描述顧客交易金額轉變趨向，為本研究的最主要貢獻。

In the previous researches, the RFM stochastic model usually combined recency (time of most recent purchase) and frequency (number of prior purchases) to estimate customer lifetime value. There are less studies only focus on monetary index to count customer contributions. This paper proposes a monetary perdition based on Markov transition matrix. First, we classify the monetary of customer transactions on different levels. Secondly, according to the proportions of every transferring level on total transaction amount, we can obtain the transition probability. Finally, the matrix of transition probability can be used to forecast the probability of customer transferring his transaction to another level. The results indicate that the normal customers show more tendency than high contribution customer to transfer their levels from low monetary amount to high monetary amount. And normal customers also have high probability of staying in the same status than the customers of high contribution. This research proposes the model of Markov transition matrix instead of traditional cluster method to describe the dynamic process of customer transaction. The framework of modeling customer contribution can provide marketing managers to segment customer.