[BUG] Development Factors

[cynthiatchetagni]

Describe the bug
while calculating manually my chainladder development factors, I do not have the same results with the .ldf_

To Reproduce

import chainladder as cl

print("pandas: " + pd.version)
print("numpy: " + np.version)
print("chainladder: " + cl.version)

sinistres = pd.read_csv("sinistres.csv",sep='\t')
sinistres.head()

sinistreINC = cl.Triangle(sinistres,origin="origin",development="development",columns="values", cumulative=False)
sinistreINC = sinistreINC.grain("OYDY")
sinistreINC

dev = cl.Development().fit(sinistreINC)
dev

sinistreINC.link_ratio
sinistreINC.age_to_age.heatmap()

dev.ldf_


**Expected behavior**

```Manually CL factors
CL | 0,975521243 | 0,976456299 | 0,995125914 | 0,997399783 | 1


**Desktop (please complete the following information):**
pandas: 2.2.3
numpy: 2.0.2
chainladder: 0.8.23

[ExemplePartrat-1.xlsx](https://github.com/user-attachments/files/19008174/ExemplePartrat-1.xlsx)

[kennethshsu]

Hi @cynthiatchetagni, welcome and thanks so much for the question!

I looked at the sample data set that you posted, which is super helpful, it makes debugging so much easier.

The cl.Development() estimator by default estimates LDFs/CDFs using volume-weighted averages. If you want the estimate to reconcile to the numbers you calculated in your excel workbook, you will need to pass in an additional parameter. More details here.

Try this:

cl.Development(average = "simple").fit(sinistreINC)

Does this help? Do you get the numbers you expect now?

[cynthiatchetagni]

Hi Kenneth, Thank you so much for your answer and thank you for this amazing package! I am still exploring it and hope to be a crack on it Best regards, Cynthia Le ven. 7 mars 2025 à 02:12, Kenneth S. Hsu ***@***.***> a écrit :

…

Hi @cynthiatchetagni <https://github.com/cynthiatchetagni>, welcome and thanks so much for the question! I looked at the sample data set that you posted, which is super helpful, it makes debugging so much easier. The cl.Development() estimator by default estimates LDFs/CDFs using volume-weighted averages. If you want the estimate to reconcile to the numbers you calculated in your excel workbook, you will need to pass in an additional parameter. More details here <https://chainladder-python.readthedocs.io/en/master/library/generated/chainladder.Development.html> . Try this: cl.Development(average = "simple").fit(sinistreINC) Does this help? Do you get the numbers you expect now? — Reply to this email directly, view it on GitHub <#552 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AOT3CQCNYYBBV25TP6WSJDD2TDW7BAVCNFSM6AAAAABX7TDOG2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDOMBVGI4DSNRRGY> . You are receiving this because you were mentioned.Message ID: ***@***.***> [image: kennethshsu]*kennethshsu* left a comment (casact/chainladder-python#552) <#552 (comment)> Hi @cynthiatchetagni <https://github.com/cynthiatchetagni>, welcome and thanks so much for the question! I looked at the sample data set that you posted, which is super helpful, it makes debugging so much easier. The cl.Development() estimator by default estimates LDFs/CDFs using volume-weighted averages. If you want the estimate to reconcile to the numbers you calculated in your excel workbook, you will need to pass in an additional parameter. More details here <https://chainladder-python.readthedocs.io/en/master/library/generated/chainladder.Development.html> . Try this: cl.Development(average = "simple").fit(sinistreINC) Does this help? Do you get the numbers you expect now? — Reply to this email directly, view it on GitHub <#552 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AOT3CQCNYYBBV25TP6WSJDD2TDW7BAVCNFSM6AAAAABX7TDOG2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDOMBVGI4DSNRRGY> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- Cynthia Tchetagni "Scan globally and reinvent locally" J.Stiglitz "La pensée complexe est une pensée qui cherche à la fois à distinguer mais sans disjoindre et à relier tout en distinguant" E.Morin