Knimi Data Mining modeling and Analysis series _003_ using Knimi to do customer segmentation

Using Knimi to do customer segmentation

Sinom

20150801

Http://blog.csdn.net/shuaihj

First, test data

Need to test the data, please leave the mailbox

Second, calculate the consumption amount and the number of consumption

1. read in ( sales data. csv)

Reading column headers

2. time format conversion

Identify the order creation date column in the specified time format

3. group Plus and sum of amounts

"Sales Amount" according to customer number sums

4. field Renaming is more readable

Statistical Results

5. counting the order groups

"Sales order Number" is de-sums based on customer number

6. field Renaming is more readable

Statistical results

7. Connect to query customer's consumption amount and consumption times

Setting connection and key columns

8. Statistical Results

9. Data Flow

third, calculate how many days no consumption

1. last consumption time

Maximum order creation date based on customer number

2. How many days did you spend ?

calculate the customer's most recent consumption, from "year January 31" There are "how many days no consumption"

3. filtering useless fields

4. Statistical Results

5. Data Flow

Four, according to the sales data to the customer hierarchical clustering calculation

1. connect to query customer's consumer information

Setting connection and key columns

Query results

2. standardization before cluster computing

Set up columns and standardized algorithms that require normalization

Standardized results

3. Compute Hierarchical Clustering

Specify distance functions, connection types, and columns that participate in cluster calculations

Hierarchical clustering Results

4. Removing noise data (global)

Enlarge Hierarchical cluster diagram

Select noise points and set to noise

Filter the noise data globally

To view the data being filtered out

5. Data Flow

four, according to the sales data to the customer K-means clustering calculation

1. calculate K-means Clustering

Specify cluster parameters and the columns that participate in the cluster calculation

View Clustering Results

2. assigning data based on cluster results

(i.e. test real data based on training model)

View Clustering Results

3. Decision Tree Training

Setting decision Tree Parameters

View Training Results

4. Data Flow

Copyright NOTICE: This article for Bo Master original article, without Bo Master permission not reproduced.

Knimi Data Mining modeling and Analysis series _003_ using Knimi to do customer segmentation