search

biolab / orange3-timeseries

🍊 :chart_with_upwards_trend: Orange add-on for analyzing, visualizing, manipulating, and forecasting time series data.
Other
62 stars 41 forks source link

Orange ARIMA model #177

Closed SuperShamoo closed 2 years ago

SuperShamoo commented 3 years ago

While doing an Orange Project in school nobody in the class was able to use the ARIMA model. the VAR model worked just fine Even the instructor was able to re-create the issue after downloading the latest version of Orange and the latest TimeSeries plugin

Download Orange v3.3 -> Install TimeSeriesv0.3.11 ARIMA model does not perform, it breaks with an error: "Error fitting model: TypeError: fit() got an unexpected keyword argument "disp""

What's your environment?

kernc commented 3 years ago

Seems like statsmodels.tsa.ARIMA.fit() no longer takes these arguments: https://github.com/biolab/orange3-timeseries/blob/a9fb2ab04dffdc8c17cb4020e94a93538099c285/orangecontrib/timeseries/models.py#L305-L306 https://github.com/statsmodels/statsmodels/blob/d990e3273fc43aac35c6fbea4485b4aa087638dc/statsmodels/tsa/arima/model.py#L222-L225 Given the deprecation notice: https://github.com/statsmodels/statsmodels/blob/d990e3273fc43aac35c6fbea4485b4aa087638dc/statsmodels/tsa/arima_model.py#L5-L13 we might consider adopting SARIMAX with fairly few changes required.

janezd commented 2 years ago

Fixed in #178.