Lab 02 - Probability¶

Goal: normalize scores and compute expected workload.

Info: Raw scores to routing

Raw model outputs must be normalized to probabilities. Then apply consistent thresholds and fallback policies across queues.

Info: Expected value

Calculate total effort from predicted intent mix. If average handle time per intent is known, estimate queue composition and staffing.

In [ ]:

raw_scores = {"account_unlock": 6.0, "vpn_issue": 2.5, "security_incident": 1.5}

total = sum(raw_scores.values())
probs = {k: v / total for k, v in raw_scores.items()}
print("Probabilities:", probs)
print("Sum:", round(sum(probs.values()), 6))

raw_scores = {"account_unlock": 6.0, "vpn_issue": 2.5, "security_incident": 1.5} total = sum(raw_scores.values()) probs = {k: v / total for k, v in raw_scores.items()} print("Probabilities:", probs) print("Sum:", round(sum(probs.values()), 6))

In [ ]:

avg_handle_minutes = {
    "account_unlock": 6,
    "vpn_issue": 14,
    "security_incident": 32,
}

expected_minutes = sum(avg_handle_minutes[k] * p for k, p in probs.items())
print("Expected handle minutes per ticket:", round(expected_minutes, 2))

avg_handle_minutes = { "account_unlock": 6, "vpn_issue": 14, "security_incident": 32, } expected_minutes = sum(avg_handle_minutes[k] * p for k, p in probs.items()) print("Expected handle minutes per ticket:", round(expected_minutes, 2))

In [ ]:

threshold = 0.80
top_intent = max(probs, key=probs.get)
top_prob = probs[top_intent]
print("Top intent:", top_intent, "probability:", round(top_prob, 3))
print("Auto-route:", top_prob >= threshold)

threshold = 0.80 top_intent = max(probs, key=probs.get) top_prob = probs[top_intent] print("Top intent:", top_intent, "probability:", round(top_prob, 3)) print("Auto-route:", top_prob >= threshold)