Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/hilalguleryuz/rfm_analysis
https://github.com/hilalguleryuz/rfm_analysis
Last synced: about 1 month ago
JSON representation
- Host: GitHub
- URL: https://github.com/hilalguleryuz/rfm_analysis
- Owner: hilalguleryuz
- Created: 2024-11-05T12:54:06.000Z (2 months ago)
- Default Branch: main
- Last Pushed: 2024-11-18T12:13:33.000Z (about 2 months ago)
- Last Synced: 2024-11-18T12:48:46.906Z (about 2 months ago)
- Size: 340 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# RFM Analysis
I performed RFM analysis using the dataset below. When calculating recency, I used the latest order date, not today's date.
The dataset is taken from this link: https://www.kaggle.com/datasets/carrie1/ecommerce-data
### RECENCY:
Last purchase date for each customer:
```sql
SELECT
customer_id,
MAX(invoicedate)
FROM rfm
WHERE customer_id IS NOT NULL
GROUP BY 1;
```
#### (Current_date) - (max_invoice_date):
```sql
WITH day_difference AS (
SELECT
customer_id,
MAX(invoicedate) AS max_invoice_date
FROM rfm
GROUP BY 1
)
SELECT
customer_id,
max_invoice_date,
current_date - max_invoice_date::date
FROM day_difference;
```
#### To get the last invoice date instead of current_date:
```sql
WITH last_invoice_date AS (
SELECT
customer_id,
MAX(invoicedate)::date AS max_invoice_date
FROM rfm
GROUP BY 1
)
SELECT
customer_id,
max_invoice_date,
'2011-12-09' - max_invoice_date AS recency
FROM last_invoice_date;
```
### FREQUENCY:
For each customer's number of orders to date:
```sql
SELECT
customer_id,
COUNT(DISTINCT invoiceno)
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1;
```
### MONETARY:
For total spending by customers:
```sql
SELECT
customer_id,
SUM(ROUND((quantity * unitprice)::integer, 0)) AS total_spent
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1;
```
To merge all three tables:
```sql
WITH recency AS (
WITH tablo AS (
SELECT
customer_id,
MAX(invoicedate)::date AS max_invoice_date
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1
)
SELECT
customer_id,
max_invoice_date,
'2011-12-09' - max_invoice_date AS recency
FROM tablo
),
frequency AS (
SELECT
customer_id,
COUNT(DISTINCT invoiceno) AS frequency
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1
),
monetary AS (
SELECT
customer_id,
SUM(ROUND((quantity * unitprice)::integer, 0)) AS monetary
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1
)
SELECT
r.customer_id,
recency,
frequency,
monetary
FROM recency r
JOIN frequency f ON r.customer_id = f.customer_id
JOIN monetary m ON r.customer_id = m.customer_id;
```
#### SQL OUTPUT
### FREQUENCY REVIEW:
How many people shopped with a certain frequency?
```sql
WITH tablo AS (
SELECT
customer_id,
COUNT(DISTINCT invoiceno) AS frequency
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1
)
SELECT
frequency,
COUNT(1)
FROM tablo
GROUP BY 1
ORDER BY 1;
```
#### SQL OUTPUT
As can be understood from the table and graph above, the number of people who shopped only once is 1494, which makes up 34% of the total number. Therefore, I gave a frequency_score of 1 to those who shopped once. Again, since the number of people who shopped twice was quite high, I gave them a score of 2. Since the percentage of people who ordered 3 and 4 times was close to the percentage of people who ordered twice, I gave them a score of 3. As a result, as can be understood from the excel table above, I grouped the shopping frequencies so that their percentages were close to each other and gave each of them a score between 1-5 accordingly. I did this grouping with CASE WHEN in the query below.
### Customer segment with Ntile & Case when:
```sql
WITH recency AS (
WITH tablo AS (
SELECT
customer_id,
MAX(invoicedate)::date AS max_invoice_date
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1
)
SELECT
customer_id,
max_invoice_date,
'2011-12-09' - max_invoice_date AS recency
FROM tablo
),
frequency AS (
SELECT
customer_id,
COUNT(DISTINCT invoiceno) AS frequency
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1
),
monetary AS (
SELECT
customer_id,
SUM(ROUND((quantity * unitprice)::integer, 0)) AS monetary
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1
)
SELECT
r.customer_id,
recency,
NTILE(5) OVER (ORDER BY recency DESC) AS recency_score,
frequency,
CASE
WHEN frequency = 1 THEN 1
WHEN frequency = 2 THEN 2
WHEN frequency = 3 OR frequency = 4 THEN 3
WHEN frequency BETWEEN 5 AND 13 THEN 4
ELSE 5
END AS frequency_score,
monetary,
NTILE(5) OVER (ORDER BY monetary) AS monetary_score
FROM recency r
JOIN frequency f ON r.customer_id = f.customer_id
JOIN monetary m ON r.customer_id = m.customer_id;
```
### Final stage:
```sql
WITH recency AS (
WITH tablo AS (
SELECT
customer_id,
MAX(invoicedate)::date AS max_invoice_date
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1
)
SELECT
customer_id,
max_invoice_date,
'2011-12-09' - max_invoice_date AS recency
FROM tablo
),
frequency AS (
SELECT
customer_id,
COUNT(DISTINCT invoiceno) AS frequency
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1
),
monetary AS (
SELECT
customer_id,
SUM(ROUND((quantity * unitprice)::integer, 0)) AS monetary
FROM rfm
WHERE invoiceno NOT LIKE 'C%'
AND customer_id IS NOT NULL
GROUP BY 1
),
scores AS (
SELECT
r.customer_id,
recency,
NTILE(5) OVER (ORDER BY recency DESC) AS recency_score,
frequency,
CASE
WHEN frequency = 1 THEN 1
WHEN frequency = 2 THEN 2
WHEN frequency = 3 OR frequency = 4 THEN 3
WHEN frequency BETWEEN 5 AND 13 THEN 4
ELSE 5
END AS frequency_score,
monetary,
NTILE(5) OVER (ORDER BY monetary) AS monetary_score
FROM recency r
JOIN frequency f ON r.customer_id = f.customer_id
JOIN monetary m ON r.customer_id = m.customer_id
)
SELECT
customer_id,
recency_score || '-' || frequency_score || '-' || monetary_score AS rfm_Score
FROM scores;
```
#### SQL OUTPUT
