Rolling

Rolling window operations

Regular

source

rolling_mean

 rolling_mean (input_array:numpy.ndarray, window_size:int,
               min_samples:Optional[int]=None)

Compute the rolling_mean over the last non-na window_size samples of the input array starting at min_samples.

Type Default Details
input_array ndarray Input array
window_size int Size of the sliding window
min_samples Optional None Minimum number of samples to produce a result, if None then it’s set to window_size
Returns ndarray Array with rolling computation

source

rolling_std

 rolling_std (input_array:numpy.ndarray, window_size:int,
              min_samples:Optional[int]=None)

Compute the rolling_std over the last non-na window_size samples of the input array starting at min_samples.

Type Default Details
input_array ndarray Input array
window_size int Size of the sliding window
min_samples Optional None Minimum number of samples to produce a result, if None then it’s set to window_size
Returns ndarray Array with rolling computation

source

rolling_max

 rolling_max (input_array:numpy.ndarray, window_size:int,
              min_samples:Optional[int]=None)

Compute the rolling_max over the last non-na window_size samples of the input array starting at min_samples.

Type Default Details
input_array ndarray Input array
window_size int Size of the sliding window
min_samples Optional None Minimum number of samples to produce a result, if None then it’s set to window_size
Returns ndarray Array with rolling computation

source

rolling_min

 rolling_min (x:numpy.ndarray, window_size:int,
              min_samples:Optional[int]=None)

Compute the rolling_min over the last non-na window_size samples of the input array starting at min_samples.

Type Default Details
x ndarray
window_size int Size of the sliding window
min_samples Optional None Minimum number of samples to produce a result, if None then it’s set to window_size
Returns ndarray Array with rolling computation

source

rolling_correlation

 rolling_correlation (x:numpy.ndarray, window_size:int)

Calculates the rolling correlation of a time series.

Type Details
x ndarray Array of time series data.
window_size int Size of the sliding window.
Returns ndarray Array with the rolling correlation for each point in time.

source

rolling_cv

 rolling_cv (x:numpy.ndarray, window_size:int)

Calculates the rolling coefficient of variation (CV) over a specified window.

Type Details
x ndarray Array of time series data.
window_size int Size of the sliding window.
Returns ndarray An array with the rolling CV for each point in time.

source

rolling_mean_positive_only

 rolling_mean_positive_only (x:numpy.ndarray, window_size:int)

Calculates the rolling mean considering only positive sales days, ignoring effects of zero demand.

Type Details
x ndarray Array of sales data.
window_size int Size of the sliding window.
Returns ndarray An array with the rolling mean for each point in time, considering only days with positive sales.

source

rolling_kurtosis

 rolling_kurtosis (x:numpy.ndarray, window_size:int)

Calculates the rolling kurtosis, helping identify the presence of outliers in sales and how data deviates from a normal distribution.

Type Details
x ndarray Array of sales data.
window_size int Size of the sliding window.
Returns ndarray Array with the rolling kurtosis for each point in time.

source

rolling_average_days_with_sales

 rolling_average_days_with_sales (x:numpy.ndarray, window_size:int)

Calculates the average number of days with sales over a window. Useful for understanding the sales frequency of each SKU.

Type Details
x ndarray Array of sales data.
window_size int Size of the sliding window.
Returns ndarray Array with the average number of days with sales for each point in time.

Seasonal

source

seasonal_rolling_mean

 seasonal_rolling_mean (input_array:numpy.ndarray, season_length:int,
                        window_size:int, min_samples:Optional[int]=None)

Compute the seasonal_rolling_mean over the last non-na window_size samples for each seasonal period of the input array starting at min_samples.

Type Default Details
input_array ndarray Input array
season_length int Length of the seasonal period
window_size int Size of the sliding window
min_samples Optional None Minimum number of samples to produce a result, if None then it’s set to window_size
Returns ndarray Array with rolling computation

source

seasonal_rolling_std

 seasonal_rolling_std (input_array:numpy.ndarray, season_length:int,
                       window_size:int, min_samples:Optional[int]=None)

Compute the seasonal_rolling_std over the last non-na window_size samples for each seasonal period of the input array starting at min_samples.

Type Default Details
input_array ndarray Input array
season_length int Length of the seasonal period
window_size int Size of the sliding window
min_samples Optional None Minimum number of samples to produce a result, if None then it’s set to window_size
Returns ndarray Array with rolling computation

source

seasonal_rolling_max

 seasonal_rolling_max (input_array:numpy.ndarray, season_length:int,
                       window_size:int, min_samples:Optional[int]=None)

Compute the seasonal_rolling_max over the last non-na window_size samples for each seasonal period of the input array starting at min_samples.

Type Default Details
input_array ndarray Input array
season_length int Length of the seasonal period
window_size int Size of the sliding window
min_samples Optional None Minimum number of samples to produce a result, if None then it’s set to window_size
Returns ndarray Array with rolling computation

source

seasonal_rolling_min

 seasonal_rolling_min (x:numpy.ndarray, season_length:int,
                       window_size:int, min_samples:Optional[int]=None)

Compute the seasonal_rolling_min over the last non-na window_size samples for each seasonal period of the input array starting at min_samples.

Type Default Details
x ndarray
season_length int Length of the seasonal period
window_size int Size of the sliding window
min_samples Optional None Minimum number of samples to produce a result, if None then it’s set to window_size
Returns ndarray Array with rolling computation