Saving a model after a warm start

I am using the below code to fit a model using a warm start. which calls the below function to write the model to file. I am able to save the initial model without any issues. However, when I come to try and save a model after a warm start (that is essentially two models that have been combined) I get the below error:

Object of type ndarray is not JSON serializable from the json.dump(model_to_json(model) code.

Can anyone help, this is quite urgent. Thank you!

` m_current = read_model_from_cloud(userId, containerId, tagId)

    m = Prophet.Prophet(daily_seasonality='auto',
                        yearly_seasonality='auto',
                        weekly_seasonality='auto',
                        interval_width=interval_width,
                        changepoint_range=0.8)
    if m_current is not None:
        m = m.fit(df, init=stan_init(m_current))
    else:
        m = m.fit(df)
    upload_model_to_cloud(m,
                          userId,
                          containerId,
                          tagId,
                          lastRun)`

with open(model_name, 'w') as fout: json.dump(model_to_json(model), fout)

The issue seems to be with the model_to_json function

[bletham]

This is fixed in 29f1417.

test code:

from fbprophet.serialize import model_to_json, model_from_json
from fbprophet import Prophet
import pandas as pd


def stan_init(m):
    res = {}
    for pname in ['k', 'm', 'sigma_obs']:
        res[pname] = m.params[pname][0][0]
    for pname in ['delta', 'beta']:
        res[pname] = m.params[pname][0]
    return res

df = pd.read_csv('../examples/example_wp_log_peyton_manning.csv')
df1 = df.loc[df['ds'] < '2016-01-19', :]  # All data except the last day
m1 = Prophet().fit(df1)
m2 = Prophet().fit(df, init=stan_init(m1)) 

mj1 = model_to_json(m1)  # works
mj2 = model_to_json(m2)  # failed before, now works

[Fasturtle1]

it should be the case that any numpy arrays are being converted to lists before serialization.

Using the debugger, I found two fields “delta” and “beta” which remain in the ndarray format. Convert them to a list format manually before calling the “model_to_json” method.

model.fit_kwargs['init']['delta'] = model.fit_kwargs['init']['delta'].tolist()
model.fit_kwargs['init']['beta'] = model.fit_kwargs['init']['beta'].tolist()