ခဏလေး ဒီလိုပဲ စဉ်းစားကြည့်
ဒီ round က round ၁ ထက် အဆင့်မြင့်ပါတယ် — skill တစ်ခုချင်းစီကို ခွဲစမ်းမယ့်အစား တစ်ခါတည်း ပေါင်းသုံးရမှာပါ။ Problem requirement ကို ဖတ်ပြီး algorithm ကို ရွေးချယ်ခြင်း, train/test accuracy gap ကနေ overfitting/underfitting ကို diagnose ခြင်း, ML project ကို scope/plan ရေးဆွဲခြင်း — ဒါတွေ ပါဝင်ပါတယ်။ Task တစ်ခုကို ၁၀ မိနစ်ခန့် ယူပါ.
လက်တွေ့ scenario နဲ့ ချိတ်ကြည့်မယ်
Task 1: 'Interpretability (ဘာကြောင့် ဒီ decision ချသလဲ ရှင်းပြနိုင်ရမယ်) က business requirement' ဆိုတဲ့ loan approval project အတွက် Decision Tree, Random Forest, Neural Network — ဘယ်ဟာ ရွေးချယ်သင့်လဲ ဆုံးဖြတ်ပြီး ဘာကြောင့်လဲ ရှင်းပါ။ Task 2: Model training accuracy 95%, test accuracy 60% ဆိုတဲ့ result ကို ဖတ်ပြီး, ဘာဖြစ်နေလဲ (overfitting) ဆုံးဖြတ်ပြီး, ဖြေရှင်းနည်း ၂ ခု အနည်းဆုံး ရေးပါ။ Task 3: E-commerce company တစ်ခုအတွက် 'ဘယ် product recommendation ကို customer ကြိုက်နှစ်သက်နိုင်လဲ' ဆိုတဲ့ project ကို Supervised ဒါမှမဟုတ် Unsupervised Learning နဲ့ approach လုပ်မလဲ ဆုံးဖြတ်ပြီး ရှင်းပါ။ Task 4: ML project တစ်ခု start လုပ်ခင် stakeholder ဆီ မေးသင့်တဲ့ question ၃ ခု (data availability, success metric, deployment plan အကြောင်း) ကို ကိုယ်တိုင် ရေးဆွဲကြည့်ပါ။
အတူတူ ကြည့်မယ်
# Task 1 - interpretability-first choice
Requirement: must explain WHY a loan was denied
-> Decision Tree (single tree, human-readable path)
-> NOT Neural Network (black box, hard to explain to
a regulator or a rejected applicant)
-> Random Forest is a middle ground (feature_importances_
helps, but individual decisions are harder to trace
than a single tree)
# Task 2 - overfitting fixes
1. Add regularization (Ridge/Lasso, or reduce max_depth
for trees)
2. Get more training data, or simplify the feature set
(remove noisy/irrelevant features)Algorithm ရွေးချယ်မှု ဆုံးဖြတ်ချက်, overfitting diagnosis/fix, ML project scoping question များ ရရှိလာမည်။၅ မိနစ် စမ်းကြည့်
Task 4 ရဲ့ stakeholder question list ကို Project 2 (Customer Churn) scenario နဲ့ ချိတ်ဆက်ပြီး, ကိုယ်တိုင် mock interview အနေနဲ့ ဖြေဆိုကြည့်ပါ (self Q&A)။
သတိလေးတစ်ချက်
'Best algorithm' ဆိုတာ universal မရှိပါ — dataset characteristic, business requirement, deployment constraint အလိုက် trade-off ရှိတယ်ဆိုတာကို ML project တိုင်းမှာ သတိရပါ.