Median of SampleForecast

[alippai]

Description

If there are even number of samples, the .median calls quantile(0.5) and that doesn't return the average (like numpy)

ChatGPT suggests the following:

def quantile(self, q: float | str) -> np.ndarray:
    q = Quantile.parse(q).value  # Assuming this returns a float between 0 and 1
    if not 0 <= q <= 1:
        raise ValueError("q should be between 0 and 1")
    
    # Compute the exact position
    pos = q * (self.num_samples - 1)
    
    # Get the integer part of the position
    lower_idx = int(pos)
    
    # Compute the weight for interpolation
    weight = pos - lower_idx
    
    # Handle the edge case where pos is the last index
    if lower_idx + 1 < self.num_samples:
        # Vectorized interpolation
        return self._sorted_samples[lower_idx] * (1 - weight) + self._sorted_samples[lower_idx + 1] * weight
    else:
        # If pos is exactly the last index, return the last sample
        return self._sorted_samples[lower_idx]
        ```