Negative dimensions are not allowed

[michael-lambe]

I am feeding the following dataset into auto_arima

Month,0
2016-01-01,129.97783044109778
2016-02-01,306.55148688938147
2016-03-01,143.46609586423057
2016-04-01,385.0286675330632
2016-05-01,80.92959253879673
2016-06-01,1058.2157327421448
2016-07-01,1247.051448666004
2016-08-01,1833.1778915985017
2016-09-01,3338.9587951991443
2016-10-01,2855.8336518614783
2016-11-01,3309.5298524577643
2016-12-01,1351.2789542083938
2017-01-01,1920.2101811761734
2017-02-01,2168.912102232124
2017-03-01,3910.982302744965
2017-04-01,3190.3251082433057
2017-05-01,1374.2227079742736
2017-06-01,1403.1415360040357
2017-07-01,953.1645718609441
2017-08-01,1413.5523140947494
2017-09-01,2821.320862583547
2017-10-01,2467.3544074992637
2017-11-01,2976.3257808230696
2017-12-01,2918.4881247635467
2018-01-01,1980.0
2018-02-01,3962.0
2018-03-01,6944.0
2018-04-01,2720.0
2018-05-01,3172.0
2018-06-01,3877.0
2018-07-01,5234.0
2018-08-01,4493.0
2018-09-01,9407.0
2018-10-01,9079.0
2018-11-01,10435.0
2018-12-01,4934.0
2019-01-01,4598.0
2019-02-01,7364.0
2019-03-01,10836.0
2019-04-01,8119.0
2019-05-01,10854.0
2019-06-01,5149.256744318752
2019-07-01,6820.377809726632
2019-08-01,9176.990725800295
2019-09-01,15991.129595953533
2019-10-01,14868.559905791291

My code is the following

import pandas as pd
from pathlib import Path
import pmdarima as pm

if __name__ == "__main__":
    y = pd.read_csv(Path(Path(__file__).parents[1], "out/" + "data.csv"), index_col="Month")
    model = pm.auto_arima(
        y,
        error_action="warn",
        seasonal=True,
        m=12,
        alpha=0.05,
        suppress_warnings=True,
        trace=True
    )  # m = 12 -> monthly
    # print(model.aic())
    model.fit(y)
    future_forecast = model.predict(n_periods=12)

I am getting the following error

ValueError: negative dimensions are not allowed
From the numeric.py file from numpy core (line 215)

I am running version pmdarima==1.3.0

I am not sure exactly what is the problem. Can you help?

[tgsmith61591]

I can replicate this, and it's pretty bizarre. Let me look into what's causing it...

[tgsmith61591]

I've got a fix for this in #193