The purpose of writing this article is to summarize the application of RMF model and AHP in product data analysis. If there is anything wrong, please leave a message in the evaluation area to discuss it.

## 1. RMF model

**RMF model**, is an important tool to measure customer value and customer profitability by extracting three indicators: R (latest consumption time), f (frequency consumption frequency) and m (monetary message amount) from customer order records.

After RMF analysis, customers can be divided into the following 8 types

**Give an example of how to calculate**

**1) . obtain original order data**

**2) Calculate the values of R, m and F**

** **Excel performs pivoting processing on the original order data. You can use the user ID as the line label, count and count the order quantity to obtain the {F value, and sum the order amount to obtain the m value. How to obtain the R value?

For the calculation of R value, to calculate the latest order time comparison, a reference time needs to be introduced. Generally, the time of data is taken as the reference time,

R = reference time – latest order time.

**3) . normalization**

Idea: calculate the extreme values of R, F and m, and then compare the R, F and m in each column with their corresponding extreme values. It is greater than the extreme value of 1 and less than the extreme value of 0. So how to calculate the extreme value? The calculation formula r extreme value = (max (R) – min (R)) / 3, and the calculation methods of F and M extreme values are the same.

**4) . layer users according to R1, M1 and F1**

** **Users are divided into eight types of customer groups according to the values of R1, M1 and F1, and different marketing strategies are carried out for different groups.

### 2. AHP hierarchy model

** AHP**It is a simple, flexible and practical multi criteria decision-making method for quantitative analysis of qualitative problems. The basic idea of analytic hierarchy process is consistent with people’s thinking on a complex decision-making problem. If you want to travel on May Day, there are three tourist attractions a, B and C for you to choose from. You will repeatedly compare the three candidate locations according to some criteria such as scenery, cost and residence, diet and travel conditions. First, you will determine the proportion of these criteria in your mind. Second, you will rank the three locations according to each criterion, Finally choose the best place.

#### An example is given to introduce the steps of analysis and operation

**1) Establish AHP hierarchy model**

When we are buying bags and making a decision about which to buy, we need to build a three-tier structure model.

This is a commonly used structure. It is analyzed from top to bottom. First, analyze the criteria layer, that is, the factors affecting the target, and the close relationship between each index and the target layer to obtain a one-dimensional weight vector. Then, we analyze the close relationship between each index of the scheme layer and each index of the criteria layer, and get as many weight vectors as there are indicators of the criteria layer. Our goal must be to analyze the closeness of the relationship between the scheme layer and the target layer, and then select a scheme with the greatest closeness, which is the final result.

**2) Construct contrast matrix**

Each criterion is compared in pairs and scored 1,3,5,7,9 according to the impact on the goal

Among them, the relative influence of two factors is between the use of the above two adjacent levels, 2,4,6,8. The integer value range is 1-9, and the score is 1 / 9 – 1

**3) Consistency inspection**

Why do consistency tests? The main reason is that people’s judgments are difficult to maintain complete consistency. In order to make the comparison of the importance of impact indicators have logical consistency, consistency test should be carried out.

In step 2, a 5th order square matrix is constructed, which is defined as a for the moment, (1) the maximum eigenvalue of matrix A is obtained, the consistency index CI is obtained by using the following formula, and the random consistency index RI is calculated

**Conformance inspection:**The consistency test of the matrix can be easily carried out by using the numerical table of consistency index, consistency ratio < 0.1 and random consistency index. The above Cr is called the consistency test formula of hierarchical single sorting. Single sorting involves only two layers, and generally involves three layers, which is called total sorting.

When CR < 0.1, it is considered that the hierarchical total ranking passes the consistency test, but it is still unable to decide which package to buy. Next, the final decision is made according to the hierarchical total ranking of the lowest level (decision-making level).

**Summary**, in many cases, RFM model and AHP hierarchy model will be used together. For example, the criteria layer of the target may be r, F and m. I have just learned this content. I have only conducted an actual battle with the relevant data of my own products. If there are any defects, please comment in the message area and we will discuss them together.