Modeling Sprint: optimize inference latency #12

Code Example

import numpy as np

def clean_series(values: list[float]):
  arr = np.array(values, dtype=float)
  median = float(np.median(arr))
  mad = float(np.median(np.abs(arr - median)))
  threshold = median + 3 * max(mad, 1e-6)
  filtered = arr[arr <= threshold]
  return {
    "median": median,
    "mad": mad,
    "count_before": len(arr),
    "count_after": len(filtered),
  }

series = [12, 13, 11, 14, 500, 12, 13, 11]
print("inspect:", "plt.plot()")
print(clean_series(series))

Commands & References

• python script.py
• df.describe()
• plt.plot()

Lab Steps

1. Prepare environment using: python script.py
2. Load a small sample dataset and validate schema.
3. Run the core code workflow and collect metrics.
4. Compare results and write one improvement note.

Exercises

• Change one hyperparameter and compare impact.
• Add one validation rule to reduce bad inputs.
• Document one failure mode and mitigation.