TVP example fails

[fonnesbeck]

Running the example associated with the tvp entry in the API docs fails:

AttributeError                            Traceback (most recent call last)
Cell In[1], line 12
      9 sales = np.random.normal(100, 10, size=len(dates))
     11 # Infer time index
---> 12 time_index = infer_time_index(dates, dates, time_resolution=5)
     14 # Define model configuration
     15 model_config = {
     16     "sales_tvp_config": {
     17         "m": 200,
   (...)
     23     }
     24 }

File ~/labs/pymc-marketing/pymc_marketing/mmm/tvp.py:261, in infer_time_index(date_series_new, date_series, time_resolution)
    237 def infer_time_index(
    238     date_series_new: pd.Series,
    239     date_series: pd.Series,
    240     time_resolution: int,
    241 ) -> npt.NDArray[np.int_]:
    242     """Infer the time-index given a new dataset.
    243 
    244     Infers the time-indices by calculating the number of days since the first date in the dataset.
   (...)
    259 
    260     """
--> 261     return (date_series_new - date_series.iloc[0]).dt.days.values // time_resolution

AttributeError: 'DatetimeIndex' object has no attribute 'iloc'

Environment details:
Python version : 3.10.16
pymc_marketing: 0.11.1
pymc : 5.21.0
arviz : 0.20.0
numpy : 1.26.4
pandas : 2.2.3

[wd60622]

I cannot find this in the notebook. Can you link? Is it latest docs? or stable?

[fonnesbeck]

https://www.pymc-marketing.io/en/stable/api/generated/pymc_marketing.mmm.tvp.html

[wd60622]

I'm making a PR for this. That logic is old

I would recommend using one of the classes here:
https://www.pymc-marketing.io/en/stable/api/generated/pymc_marketing.mmm.hsgp.html

For instance, HSGP: https://www.pymc-marketing.io/en/stable/api/generated/pymc_marketing.mmm.hsgp.HSGP.html#pymc_marketing.mmm.hsgp.HSGP